Computational tools for enzyme engineering: bridging the gap between enzymologists and expert simulation
Lead Research Organisation:
University of Bristol
Department Name: Chemistry
Abstract
It is becoming increasingly popular to use the powerful principles present in nature to our advantage. A key example is the extraordinary ability of organisms to make molecules with high specificity (pure, potentially complex molecules are obtained) and efficiency (little energy is used). Nature uses enzymes, proteins that act as catalysts, to achieve this. These enzymes typically work under mild conditions. Enzymes are already used in industry to make molecules that we require in cost-efficient, comparatively green and sustainable processes. However, nature has not provided us with an enzyme to suit the production of every desired molecule; typically enzymes only catalyze specific chemical reactions with specific starting materials. But the process of evolution teaches us that enzymes may be malleable for engineering different properties. For example, making small changes (mutations) in specific amino acids (the building blocks of proteins) of enzymes can allow these enzymes to accept different substrates and thereby catalyze the formation of new, desired molecules. Even though it is possible to determine the positions of atoms in an enzyme with great detail (e.g. using X-ray crystallography), the full effects of making changes to amino acids are not evident. This limits researchers in assessing what the (beneficial or non-beneficial) effects of such mutations are. It is possible to predict these effects with sophisticated computer simulation methods, but performing the necessary simulations requires expert knowledge. The researchers that are involved in optimizing enzymes to obtain new catalysts for making desired molecules are therefore usually limited to guessing what the effect of mutations is based on static structures alone. To bridge the gap between such experimental researchers and those that are experts in computer simulation, we aim to make expert simulation methods available through an interface that is familiar to the experimental researchers. Our project will involve the development of simulation protocols that assess the effects of mutations, using state-of-the-art methods that include molecular dynamics simulations and quantum chemistry calculations. The protocols will be designed such that they can be run on standard computers and they will be made accessible through an easy and familiar interface for experimental researchers (without the need for in-depth training in computer simulation). In addition, the protocols will allow high-throughput screening of 100s of mutations on high-performance computer clusters. The end result will be that researchers not skilled in computer simulation can easily assess the potential influence of mutations using their own computers, and that high-throughput screening of enzyme variants can be performed computationally. This can potentially save a lot of time and resources in the process of adapting an enzyme for a desired reaction. In addition, collaboration between researchers with complementary skills will be encouraged. The tools developed will also be beneficial in related fields, for example in designing effective drugs and understanding inheritable diseases.
Technical Summary
To obtain efficient biocatalysts, re-design or 'de novo' design of enzymes is often needed. Computational methods are increasingly employed for this purpose, but general and efficient workflows to assess how mutations affect enzymes are not yet available and accessible to experimentalists. The aim of this project is therefore to provide computational tools to a) run state-of-the-art simulation to assess the influence of mutations by non-experts on standard computers and b) perform computational high-throughput screening of enzyme variants on HPC facilities. The tools will be geared towards enzyme (re)design, but will also be useful more generally for structural and synthetic biology, drug design and health research. The protocols will go significantly beyond what is currently accessible to experimentalists, by assessing effects on the structure, dynamics, electronic properties and catalytic efficiency of enzymes. To ensure a broad uptake of the tools, we will actively develop and engage with the user community.
The project will deliver three main outputs:
1) Computationally inexpensive protocols for use on standard computers and HPC facilities. The protocols will use state-of-the-art simulation techniques, in particular molecular dynamics simulation and quantum mechanical / molecular mechanical (QM/MM) calculations, and provide relevant analysis of the simulation data.
2) A core software program that links together simulation setup and driving of software for MM and QM calculations, to seamlessly run the protocols mentioned above. The program will be set up such that interfaces to different external software packages can be written easily. The software will be released under an open-source (GPL) licence.
3) User-friendly plugins to molecular visualization software popular with structural biologists. With the plugins, released as open-source (GPL licence), users can run the core software program and then display the results in the molecular viewer.
The project will deliver three main outputs:
1) Computationally inexpensive protocols for use on standard computers and HPC facilities. The protocols will use state-of-the-art simulation techniques, in particular molecular dynamics simulation and quantum mechanical / molecular mechanical (QM/MM) calculations, and provide relevant analysis of the simulation data.
2) A core software program that links together simulation setup and driving of software for MM and QM calculations, to seamlessly run the protocols mentioned above. The program will be set up such that interfaces to different external software packages can be written easily. The software will be released under an open-source (GPL) licence.
3) User-friendly plugins to molecular visualization software popular with structural biologists. With the plugins, released as open-source (GPL licence), users can run the core software program and then display the results in the molecular viewer.
Planned Impact
This project will lead to new, easy-to-use and intuitive software designed to suit experimental researchers with no or little expertise in computational modelling and simulation. The software will thus facilitate basic and applied biochemical research by helping experimental scientists better understand and predict the effect of mutations. The software, related protocols and demonstration cases are primarily aimed at those interested in (re)designing enzymes for use as biocatalysts; this is an area of research that is also actively being pursued in industry, due to its ability to enhance the cost-effectiveness and sustainability of synthesis of chemical compounds and production of biofuels, for example. The tools produced are therefore expected to be taken up in industry as well as academia.
In the short term, the proposed project will deliver tools to help increase understanding and successful engineering of enzymes. The same tools will similarly help related biochemical research, such as protein and drug design. Significant efforts will be made to introduce interested experimental researchers to the tools developed, and train them in their use. By giving researchers with primarily experimental expertise access to and explanation of expert simulation methods, the skill-sets of these researchers will be increased. Outreach to experimental researchers will also encourage further cross-disciplinary research and related transfer of knowledge and skills. In addition, the researcher responsible for software development will obtain valuable programming experience that will be useful in his future academic or non-academic career.
In the medium term, the protocols and software that will be developed as part of the project (especially those for computational high-throughput screening of enzyme variants) should significantly reduce the time and resources spent to optimize biocatalysts for particular reactions. This has direct economic and environmental benefits (biocatalysts can be brought to market sooner, with reduced use of chemicals and energy in the development process). Similarly, computational assessment of the effect of mutations may avoid spending on producing and testing protein variants or drug lead compounds that subsequently prove ineffective.
In the long term, the commercial availability of biocatalysts optimized with help of the developed tools can potentially have enormous economic and environmental benefits by allowing the sustainable production of desired molecules, including fine-chemicals, drugs and biofuels. Biocatalysts are already starting to transform our current chemical industry by improvements in the methods, cost-effectiveness, safety, health, and environmental impact of the processes involved, and these impacts will be extended by the availability of additional biocatalysts. In addition, the application of the tools to help with and speed up drug design, for example to develop anti-viral and antibiotic compounds active against species that are currently drug-resistant, has obvious significant benefits for health in addition to the economic benefits. Similarly, the tools should contribute in helping to fulfil the enormous potential of synthetic biology to deliver societal and economic benefits.
In the short term, the proposed project will deliver tools to help increase understanding and successful engineering of enzymes. The same tools will similarly help related biochemical research, such as protein and drug design. Significant efforts will be made to introduce interested experimental researchers to the tools developed, and train them in their use. By giving researchers with primarily experimental expertise access to and explanation of expert simulation methods, the skill-sets of these researchers will be increased. Outreach to experimental researchers will also encourage further cross-disciplinary research and related transfer of knowledge and skills. In addition, the researcher responsible for software development will obtain valuable programming experience that will be useful in his future academic or non-academic career.
In the medium term, the protocols and software that will be developed as part of the project (especially those for computational high-throughput screening of enzyme variants) should significantly reduce the time and resources spent to optimize biocatalysts for particular reactions. This has direct economic and environmental benefits (biocatalysts can be brought to market sooner, with reduced use of chemicals and energy in the development process). Similarly, computational assessment of the effect of mutations may avoid spending on producing and testing protein variants or drug lead compounds that subsequently prove ineffective.
In the long term, the commercial availability of biocatalysts optimized with help of the developed tools can potentially have enormous economic and environmental benefits by allowing the sustainable production of desired molecules, including fine-chemicals, drugs and biofuels. Biocatalysts are already starting to transform our current chemical industry by improvements in the methods, cost-effectiveness, safety, health, and environmental impact of the processes involved, and these impacts will be extended by the availability of additional biocatalysts. In addition, the application of the tools to help with and speed up drug design, for example to develop anti-viral and antibiotic compounds active against species that are currently drug-resistant, has obvious significant benefits for health in addition to the economic benefits. Similarly, the tools should contribute in helping to fulfil the enormous potential of synthetic biology to deliver societal and economic benefits.
Publications
Abramiuk M
(2024)
Biocatalytic pathways, cascades, cells and systems: general discussion.
in Faraday discussions
Acevedo-Rocha C
(2024)
Artificial, biomimetic and hybrid enzymes: general discussion.
in Faraday discussions
Acevedo-Rocha C
(2018)
P450-Catalyzed Regio- and Diastereoselective Steroid Hydroxylation: Efficient Directed Evolution Enabled by Mutability Landscaping
in ACS Catalysis
Acevedo-Rocha C
(2024)
Enzyme evolution, engineering and design: mechanism and dynamics: general discussion.
in Faraday discussions
Ainsley J
(2018)
Structural Insights from Molecular Dynamics Simulations of Tryptophan 7-Halogenase and Tryptophan 5-Halogenase.
in ACS omega
Ainsley J
(2018)
Combined Quantum Mechanics and Molecular Mechanics Studies of Enzymatic Reaction Mechanisms.
in Advances in protein chemistry and structural biology
Aldeghi M
(2017)
Statistical Analysis on the Performance of Molecular Mechanics Poisson-Boltzmann Surface Area versus Absolute Binding Free Energy Calculations: Bromodomains as a Case Study
in Journal of Chemical Information and Modeling
Ali HS
(2019)
Entropy of Simulated Liquids Using Multiscale Cell Correlation.
in Entropy (Basel, Switzerland)
Althorpe SC
(2016)
Non-adiabatic reactions: general discussion.
in Faraday discussions
Amaro RE
(2020)
Biomolecular Simulations in the Time of COVID19, and After.
in Computing in science & engineering
| Description | We have developed computational tools to model enzymes (biological catalysts) and the reactions they catalyse. Enzymes make possible all chemical reactions required for the survival and reproduction of living systems. The remarkable power of enzymes to catalyse reactions rapidly, selectively and efficiently under mild conditions is exploited in biotechnology. We are, however, not yet taking full advantage of enzymes as biocatalysts, whereas more widespread application of biocatalysts in industry is desired. To enable the use of enzymes as biocatalysts for reactions of interest, modifications are often necessary; for example to improve efficiency or alter and broaden substrate specificity. Recent state-of-the-art examples of such modifications, or re-design, often employ a combination of knowledge-based site-specific engineering and directed evolution. De novo enzyme design, which can be used to obtain biocatalysts for non-natural chemical reactions, has emerged as a promising field in the past decade and typically uses a similar combined strategy. For both re-design and de novo design, computational methods are now increasingly playing a key role in the design strategies employed. Specifically, computational simulation methods can help to predict the influence of amino-acid mutations far beyond the static view of modelling a new side-chain into a crystal structure. Changes in the structure, electrostatics and dynamics of an enzyme that affect substrate binding and catalysis can be simulated. It is increasingly recognized that all these effects, including protein dynamics, must be considered to allow successful enzyme (re)design. Beyond the need for (computationally) assessing the effect of mutations in enzyme (re)design, the wider fields of protein and drug design also benefit from such assessment. In synthetic biology, mutations are routinely made with the aim to improve designed functions of proteins other than biocatalysis, such as protein-protein or protein-ligand complex formation. Mutations that occur in rapidly evolving systems such as viruses or bacteria are often a challenge for drug design. Computational simulation can be used to understand why certain mutations or enzyme variants cause antibiotic resistance. For the design of drugs that target such rapidly evolving systems, efficient computational prediction of the influence of mutations on drug interaction will be beneficial. Although computer simulations can, in principle, efficiently assess different effects of mutations on enzymes and other proteins (structural, dynamical and catalytic), setting up and performing such simulations currently requires significant knowledge of and experience with biomolecular simulation methods and a range of different programs. Each of these programs comes with its own requirements for installation, input preparation, usage and analysis of results. The application of combined quantum mechanics/molecular mechanics (QM/MM) methods to the study of biological systems is a thriving and growing area, with advances in computer hardware and software opening up exciting new avenues of research. The award of the 2013 Nobel Prize for Chemistry to Martin Karplus, Michael Levitt and Arieh Warshel, recognises the development of combined quantum mechanical / molecular mechanical (QM/MM) methods and the important role these methods can play in our understanding of enzyme-catalysed reactions. The Diels-Alder reaction, a [4 + 2] cycloaddition of a conjugated diene to a dienophile, is one of the most powerful reactions in synthetic chemistry. Biocatalysts capable of unlocking new and efficient Diels-Alder reactions would have major impact. Here we present a molecular-level description of the reaction mechanism of the spirotetronate cyclase AbyU, an enzyme shown here to be a bona fide natural Diels-Alderase. Using enzyme assays, X-ray crystal structures, and simulations of the reaction in the enzyme, we reveal how linear substrate chains are contorted within the AbyU active site to facilitate a transannular pericyclic reaction. This study provides compelling evidence for the existence of a natural enzyme evolved to catalyze a Diels-Alder reaction and shows how catalysis is achieved. Researchers from the School of Chemistry, and the University of Parma, Italy, have used molecular simulations to understand resistance to osimertinib - an anticancer drug used to treat types of lung cancer. Osimertinib binds tightly to a protein, epidermal growth factor receptor (EGFR), which is overexpressed in many tumours. EGFR is involved in a pathway that signals for cell proliferation, and so is a target for drugs. Blocking the action of EGFR (inhibiting it) can switch it off, and so is a good way to treat the disease. Osimertinib is an effective anticancer drug that works in this way. It is used to treat non-small-cell lung cancer (NSCLC), in cases where the cancer cells have a particular (T790M) mutant form of EGFR. It is a so-called 'third-generation' EGFR inhibitor, which was approved as a cancer treatment in 2017. Osimertinib is a covalent inhibitor: as such, it binds irreversibly to EGFR by forming a chemical bond with it. Although patients generally respond well to osimertinib, most acquire drug resistance within one year of treatment, so the drug stops working. Drug resistance arises because the EGFR protein mutates, so that the drug binds less tightly. One such mutation, called L718Q, was recently discovered in patients in the clinic by the Medical Oncology Unit of the University Hospital of Parma. In this drug resistant mutant, a single amino acid is changed. Unlike other drug resistant mutants, it was not at all clear how this change stops the drug from binding effectively, information potentially crucial in developing new drugs to overcome resistance. Now, a collaboration between medicinal and computational chemists and clinical oncologists has revealed exactly how subtle changes in the protein target cause drug resistance. Using a range of advanced molecular simulation techniques, scientists from the Universities of Bristol and Parma, Italy, showed that the structure of the mutant protein changes in a way that stops the drug reacting and binding to it. Professor Adrian Mulholland, said: "This work shows how molecular simulations can reveal mechanisms of drug resistance, which can be subtle and non-obvious. "In particular, here we've used combined quantum mechanics/molecular mechanics (QM/MM) methods, which allow us to study chemical reactions in proteins. "This is crucial in investigating covalent inhibitors, which react with their biological targets, and are the focus of growing interest in the pharmaceutical industry." His collaborators, Professor Alessio Lodola and Professor Marco Mor of the Drug Design and Discovery group at the University of Parma, added: "It was an exciting experience to work closely with clinical colleagues who identified the mutant, and to help analyse its effects. "Now the challenge is to exploit this discovery in the development of novel drugs targeting EGFR mutants for cancer treatment in future." This research is a collaboration between The Drug Design and Discovery group of the Department of Food and Drug at the University of Parma; the Medical Oncology Unit, University Hospital of Parma and the Centre for Computational Chemistry, in the School of Chemistry of the University of Bristol. Researchers at the University of Bristol are pioneering the use of virtual reality (VR) as a tool to design the next generation of drug treatments. The findings, published in the journal PLOS One describe how researchers used VR to understand how common medications work on a molecular level. Many drugs are small molecules, and discovering new drugs involves finding molecules that bind to biological targets like proteins. In the study, users were able to use VR to 'step inside' proteins and manipulate them, and the drugs binding to them, in atomic detail, using interactive molecular dynamics simulations in VR (iMD-VR). Using this iMD-VR approach, researchers 'docked' drug molecules into proteins and were able to predict accurately how the drugs bind. Among the systems studied were drugs for flu and HIV. Professor Adrian Mulholland, from the University of Bristol's Centre for Computational Chemistry, and co-lead of the work, said: "Many drugs work by binding to proteins and stopping them working. For example, by binding to a particular virus protein, a drug can stop the virus from reproducing. "To bind well, a small molecule drug needs to fit snugly in the protein. An important part of drug discovery is finding small molecules that bind tightly to specific proteins, and understanding what makes them bind tightly, which helps to design better drugs. "To design new therapies, researchers need to understand how drug molecules fit into their biological targets. To do this, we use VR to represent them as fully three-dimensional objects. Users can then fit a drug within the 'keyhole' of a protein binding site to discover how they fit together." In the study, users were set the task of binding drugs to protein targets such as influenza neuraminidase and HIV protease. Tests showed that users were able to predict correctly how the drugs bind to their protein targets. By pulling the drug into the protein, they could build structures that are very similar to the structures of the drug complexes found from experiments. Even non-experts were able to dock drugs into the proteins effectively. This shows that interactive VR can be used to predict accurately how new potential drugs bind to their targets. The study shows how VR can be used effectively in structure-based drug design, even by non-experts. It uses readily available VR equipment and an open source software framework, so can be applied by anyone. Professor Mulholland added: "An important aspect of the work is that the drugs, and their protein targets, are fully flexible: we model their structural changes and dynamics, and users can manipulate them interactively to find how drugs interact with their biological targets. This is a really exciting and powerful way to model drug binding. We have shown in this work that it gives accurate results. These tools will be useful in the design and development of new drugs." Dr David Glowacki, Royal Society Senior Research Fellow in Bristol's School of Chemistry and Department of Computer Science, said: "Our results show that it is possible to unbind and rebind drugs from protein targets on a simulation timescale significantly shorter than the timescale of similar events observed using non-interactive molecular dynamics engines. "It is also important to note that the full unbinding and rebinding events generated using iMD-VR were achieved by the users in less than five minutes of real time. "Where non-expert users had trace atoms showing them the correct pose, all participants were able to establish a docking pose which was close enough to the starting structure to be scientifically considered redocked. "Where no trace atoms were present, binding poses understandably had more variation, but users were still able to get within the same range of the accepted bound position for all three systems. These results were achieved within a single hour-long training session with each participant, demonstrating the usability of this VR framework." This research was supported by funding from EPSRC and the Royal Society. Further information Paper: 'Interactive molecular dynamics in virtual reality for accurate flexible protein-ligand docking' by H. Deeds, R. Walters, S. Hare, M. O'Connor, A. Mulholland and D. Glowacki in PLOS One Designing better enzymes: Insights from directed evolution Authors H Adrian Bunzel, JL Ross Anderson, Adrian J Mulholland Publication date 2021/4/1 Source Current Opinion in Structural Biology Volume 67 Pages 212-218 Publisher Elsevier Current Trends Description De novo enzymes can be created by computational design and directed evolution. Here, we review recent insights into the origins of catalytic power in evolved designer enzymes to pinpoint opportunities for next-generation designs: Evolution precisely organizes active sites, introduces catalytic H-bonding networks, invokes electrostatic catalysis, and creates dynamical networks embedding the active site in a reactive protein scaffold. Such insights foster our fundamental knowledge of enzyme catalysis and fuel the future design of tailor-made enzymes. |
| Exploitation Route | The computational tools developed will be useful for researchers in academia and industry working on enzymes, e.g. for biocatalysis. All software will be released using an open source, GPL license, and will be hosted on the github site. Final code will also be accessible via a Google Code repository, and made available through CCPForge (ccpforge.cse.rl.ac.uk). Google code and github are publicly accessible repositories, which can provide immediate, free access to the software while it is in the process of development. In addition, packaged releases of the software will be produced and hosted through CCPForge as well as data.bris (see above) to ensure its integrity, availability and longevity. Github and google code provide indefinite hosting and archiving of the data, together with versioning and automatic metadata generation. This will enable the software to be advertised to automatic software discovery services, linked to from third party sites and the code to be re-used by interested parties. To ensure further visibility to the user community, links to the core program and its plugins to visualization programs will be provided on the CCP-BioSim website (ccpbiosim.ac.uk) and further download links for the plugins will be placed on their respective websites. In addition, documentation and installation instructions for the PyMOL plugin will also be placed on its designated wiki site (http://pymolwiki.org/index.php/Category:Plugins). Detailed descriptions and the required input files for the developed simulation protocols (for all supported simulation software packages) will be made publicly available via a dedicated github site (https://github.com/marcvanderkamp/enzlig_tools) and (when finalised) through data.bris (Bristol's Research Data Repository). The latter allows free, secure and accessible storage (including unique Digital Object Identifiers (DOIs)) for reasonable size data sets; simulation protocols and input will require minimal storage ~1 GB. To make the protocols easily accessible, DOIs will be generated and links placed on the CCP-BioSim website (http://www.ccpbiosim.ac.uk). Designing better enzymes: Insights from directed evolution Authors H Adrian Bunzel, JL Ross Anderson, Adrian J Mulholland Publication date 2021/4/1 Source Current Opinion in Structural Biology Volume 67 Pages 212-218 Publisher Elsevier Current Trends Description De novo enzymes can be created by computational design and directed evolution. Here, we review recent insights into the origins of catalytic power in evolved designer enzymes to pinpoint opportunities for next-generation designs: Evolution precisely organizes active sites, introduces catalytic H-bonding networks, invokes electrostatic catalysis, and creates dynamical networks embedding the active site in a reactive protein scaffold. Such insights foster our fundamental knowledge of enzyme catalysis and fuel the future design of tailor-made enzymes. How enzymes - the biological proteins that act as catalysts and help complex reactions occur - are 'tuned' to work at a particular temperature is described in new research from groups in New Zealand and the UK, including the University of Bristol. Professor Vic Arcus (University of Waikato) and colleagues, including Bristol's Professor Adrian Mulholland and Dr Marc van der Kamp, showed that the heat capacity of enzymes changes during a reaction as the enzymes tighten up. Exactly how much the enzymes tighten up is the critical factor in determining the temperature at which they work best. These findings could provide a route to designing better biocatalysts for use in chemical reactions in industrial processes, such as the production of drugs. Enzymes have an optimum temperature at which they are most catalytically active. Above that temperature, they become less active. Previously, it was thought that this was because enzymes unfolded (lost their functional shape) at higher temperatures, but actually they typically become less active at higher temperatures even though they maintain their functional shape. So what makes them less active? And what is it that causes enzymes from different organisms to have different catalytic activities at the same temperature? Enzymes from organisms that live at normal temperatures are not very active at low temperatures, while cold-adapted enzymes are active in the cold - why, when they have very similar structures? The new research, published as a 'New Concept' in Biochemistry (and selected for the American Chemical Society (ACS) Editors' Choice), shows that a basic physical property - the heat capacity - explains and predicts the temperature dependence of enzymes. The heat capacity of a substance is the amount of heat required to raise its temperature by one degree. For enzymes, the heat capacity changes during the reaction and this change is 'tuned' to give the optimal temperature. Professor Mulholland said: "Our theory - macromolecular rate theory, (MMRT) - applies to all enzymes, and so will have a critical role in predicting metabolic activity as a function of temperature. "We also expect to see characteristics of MMRT at the level of cells, whole organisms and even ecosystems. This means that it is important in understanding and predicting the response of biological systems to temperature changes, for example, how ecosystems will respond to temperature changes associated with climate change." The theory also explains why enzymes are so big (the more 'difficult' the chemistry to catalyse, the bigger the enzyme). It also hints at why proteins were eventually preferred by evolution over nucleic acids as catalysts in biology: proteins offer much more ability to 'tune' dynamics and their response to chemical reactions. Paper 'On the Temperature Dependence of Enzyme-catalyzed Rates' by Vickery L. Arcus, Erica J. Prentice, Joanne K. Hobbs, Adrian J. Mulholland, Marc W. Van der Kamp, Christopher R. Pudney, Emily J. Parker and Louis A. Schipper in Biochemistry https://pubs.acs.org/doi/abs/10.1021/acs.biochem.5b01094 Researchers at the University of Bristol are pioneering the use of virtual reality (VR) as a tool to design the next generation of drug treatments. The findings, published in the journal PLOS One describe how researchers used VR to understand how common medications work on a molecular level. Many drugs are small molecules, and discovering new drugs involves finding molecules that bind to biological targets like proteins. In the study, users were able to use VR to 'step inside' proteins and manipulate them, and the drugs binding to them, in atomic detail, using interactive molecular dynamics simulations in VR (iMD-VR). Using this iMD-VR approach, researchers 'docked' drug molecules into proteins and were able to predict accurately how the drugs bind. Among the systems studied were drugs for flu and HIV. Professor Adrian Mulholland, from the University of Bristol's Centre for Computational Chemistry, and co-lead of the work, said: "Many drugs work by binding to proteins and stopping them working. For example, by binding to a particular virus protein, a drug can stop the virus from reproducing. "To bind well, a small molecule drug needs to fit snugly in the protein. An important part of drug discovery is finding small molecules that bind tightly to specific proteins, and understanding what makes them bind tightly, which helps to design better drugs. "To design new therapies, researchers need to understand how drug molecules fit into their biological targets. To do this, we use VR to represent them as fully three-dimensional objects. Users can then fit a drug within the 'keyhole' of a protein binding site to discover how they fit together." In the study, users were set the task of binding drugs to protein targets such as influenza neuraminidase and HIV protease. Tests showed that users were able to predict correctly how the drugs bind to their protein targets. By pulling the drug into the protein, they could build structures that are very similar to the structures of the drug complexes found from experiments. Even non-experts were able to dock drugs into the proteins effectively. This shows that interactive VR can be used to predict accurately how new potential drugs bind to their targets. The study shows how VR can be used effectively in structure-based drug design, even by non-experts. It uses readily available VR equipment and an open source software framework, so can be applied by anyone. Professor Mulholland added: "An important aspect of the work is that the drugs, and their protein targets, are fully flexible: we model their structural changes and dynamics, and users can manipulate them interactively to find how drugs interact with their biological targets. This is a really exciting and powerful way to model drug binding. We have shown in this work that it gives accurate results. These tools will be useful in the design and development of new drugs." Dr David Glowacki, Royal Society Senior Research Fellow in Bristol's School of Chemistry and Department of Computer Science, said: "Our results show that it is possible to unbind and rebind drugs from protein targets on a simulation timescale significantly shorter than the timescale of similar events observed using non-interactive molecular dynamics engines. "It is also important to note that the full unbinding and rebinding events generated using iMD-VR were achieved by the users in less than five minutes of real time. "Where non-expert users had trace atoms showing them the correct pose, all participants were able to establish a docking pose which was close enough to the starting structure to be scientifically considered redocked. "Where no trace atoms were present, binding poses understandably had more variation, but users were still able to get within the same range of the accepted bound position for all three systems. These results were achieved within a single hour-long training session with each participant, demonstrating the usability of this VR framework." This research was supported by funding from EPSRC and the Royal Society. Further information Paper: 'Interactive molecular dynamics in virtual reality for accurate flexible protein-ligand docking' by H. Deeds, R. Walters, S. Hare, M. O'Connor, A. Mulholland and D. Glowacki in PLOS One |
| Sectors | Chemicals Digital/Communication/Information Technologies (including Software) Education Healthcare Manufacturing including Industrial Biotechology Pharmaceuticals and Medical Biotechnology |
| URL | https://sites.google.com/site/mulhollandresearchgroup/ |
| Description | The computational tools developed for enzyme simulations have been demonstrated to industrial researchers working in biocatalysis and pharmaceuticals, and via the Catalysis Hub. Carbapenems, 'last resort' antibiotics for many bacterial infections, can now be broken down by several class A ß-lactamases (i.e. carbapenemases). Here, carbapenemase activity is predicted through QM/MM dynamics simulations of acyl-enzyme deacylation, requiring only the 3D structure of the apo-enzyme. This may |
| First Year Of Impact | 2014 |
| Sector | Agriculture, Food and Drink,Chemicals,Digital/Communication/Information Technologies (including Software),Healthcare,Pharmaceuticals and Medical Biotechnology |
| Impact Types | Economic |
| Description | Government urges funders to consider scrapping Researchfish |
| Geographic Reach | National |
| Policy Influence Type | Contribution to a national consultation/review |
| URL | https://www.gov.uk/government/publications/review-of-research-bureaucracy |
| Description | Large-Scale Computing Opportunities and Challenges. A Report from a Supercomputing Task Force for BBSRC and MRC |
| Geographic Reach | National |
| Policy Influence Type | Contribution to a national consultation/review |
| URL | https://doi.org/10.5281/zenodo.11079792 |
| Description | Research data and policy |
| Geographic Reach | National |
| Policy Influence Type | Contribution to new or improved professional practice |
| URL | https://www.chemistryworld.com/news/ukri-finds-itself-in-hot-water-too-over-researchfish-cyberbullyi... |
| Description | Science after Brexit: bright spots on the Horizon? |
| Geographic Reach | Europe |
| Policy Influence Type | Contribution to a national consultation/review |
| URL | https://digital-strategy.ec.europa.eu/en/news/united-kingdom-joins-horizon-europe-programme#:~:text=... |
| Description | UKRI research data capture approaches |
| Geographic Reach | National |
| Policy Influence Type | Contribution to new or improved professional practice |
| URL | https://www.researchprofessionalnews.com/rr-news-uk-research-councils-2023-1-researchfish-tweets-aga... |
| Description | AMR Global Development Award 2017 |
| Amount | £88,000 (GBP) |
| Funding ID | MR/R014922/1 |
| Organisation | Medical Research Council (MRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 09/2017 |
| End | 03/2018 |
| Description | BBSRC sLoLa: Innovative Routes to Monoterpene Hydrocarbons and Their High Value Derivatives |
| Amount | £3,038,984 (GBP) |
| Funding ID | BB/M000354/1 |
| Organisation | Biotechnology and Biological Sciences Research Council (BBSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 09/2010 |
| End | 09/2019 |
| Description | BEORHN: Bacterial Enzymatic Oxidation of Reactive Hydroxylamine in Nitrification via Combined Structural Biology and Molecular Simulation |
| Amount | £181,398 (GBP) |
| Funding ID | BB/V016768/1 |
| Organisation | Biotechnology and Biological Sciences Research Council (BBSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 01/2022 |
| End | 12/2024 |
| Description | Biocatalysis and Biotransformation: A 5th Theme for the National Catalysis Hub |
| Amount | £3,053,639 (GBP) |
| Funding ID | EP/M013219/1 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 01/2015 |
| End | 12/2019 |
| Description | BristolBridge has contributed to the University of Bristol being the UK's largest recipient of RCUK AMR cross-council funding awards both in number (7) and value (£5.268M). 6 related AMR grants awarded with BristolBridge PIs/Co-Is include MRC-led AMR. |
| Amount | £5,268,000 (GBP) |
| Organisation | Medical Research Council (MRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 09/2016 |
| End | 09/2019 |
| Description | CCP-BioSim: Biomolecular Simulation at the Life Sciences Interface |
| Amount | £235,706 (GBP) |
| Funding ID | EP/M022609/1 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 06/2015 |
| End | 04/2021 |
| Description | CCP-BioSim: Biomolecular simulation at the life sciences interface |
| Amount | £287,541 (GBP) |
| Funding ID | EP/J010588/1 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 09/2011 |
| End | 09/2015 |
| Description | CCPBioSim: Biomolecular Simulation at the Life Science Interface |
| Amount | £345,687 (FKP) |
| Funding ID | EP/T026308/1 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 11/2020 |
| End | 10/2025 |
| Description | Carbapenem Antibiotic Resistance in Enterobacteriaceae: Understanding Interactions of KPC Carbapenemases with Substrates and Inhibitors |
| Amount | £668,396 (GBP) |
| Funding ID | MR/T016035/1 |
| Organisation | Medical Research Council (MRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 01/2020 |
| End | 01/2023 |
| Description | Confidence in Concept 'Developing a mobile device for rapid antimicrobial resistance detection in primary care' |
| Amount | £74,685 (GBP) |
| Organisation | Medical Research Council (MRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 05/2017 |
| End | 05/2018 |
| Description | EPSRC |
| Amount | £188,950 (GBP) |
| Funding ID | E/EP/G007705/1 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 09/2013 |
| End | 03/2014 |
| Description | Inquire: Software for real-time analysis of binding |
| Amount | £105,748 (GBP) |
| Funding ID | BB/K016601/1 |
| Organisation | Biotechnology and Biological Sciences Research Council (BBSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 08/2013 |
| End | 09/2014 |
| Description | Nicotinic Ligand Development to Target Smoking Cessation and Gain a Molecular Level Understanding of Partial Agonism |
| Amount | £724,553 (GBP) |
| Funding ID | EP/N024117/1 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 05/2016 |
| End | 11/2019 |
| Description | Oracle for Research Cloud Fellowship |
| Amount | $100,000 (USD) |
| Organisation | Oracle Corporation |
| Sector | Private |
| Country | United States |
| Start | 02/2023 |
| End | 12/2023 |
| Description | PREDACTED Predictive computational models for Enzyme Dynamics, Antimicrobial resistance, Catalysis and Thermoadaptation for Evolution and Desig |
| Amount | € 2,482,332 (EUR) |
| Funding ID | 101021207 |
| Organisation | European Research Council (ERC) |
| Sector | Public |
| Country | Belgium |
| Start | 09/2021 |
| End | 09/2026 |
| Description | Synthetic Biology Research Centre. BrisSynBio: Bristol Centre for Synthetic Biology |
| Amount | £13,528,180 (GBP) |
| Funding ID | BB/L01386X/1 |
| Organisation | Biotechnology and Biological Sciences Research Council (BBSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 06/2014 |
| End | 07/2019 |
| Description | https://gtr.ukri.org/person/2A2990B1-E1E1-4888-8848-7C256C3A3B43 |
| Amount | £20,009,000 (GBP) |
| Funding ID | https://gtr.ukri.org/person/2A2990B1-E1E1-4888-8848-7C256C3A3B43 |
| Organisation | United Kingdom Research and Innovation |
| Sector | Public |
| Country | United Kingdom |
| Start | 01/2006 |
| End | 02/2033 |
| Title | An expandable, modular de novo protein platform for precision redox engineering |
| Description | Pioneering study signals new era of environment-friendly programmable bioelectronics Image shows the design of a protein nanowire, with the green arrow indicating electron flow. Ross Anderson Image shows structural analysis of the protein-based wire, comparing the model of the designed protein (shown in red) with the experimentally determined structure (in grey). Ross Anderson Press release issued: 25 July 2023 Researchers have created a unique microscopic toolkit of 'green' tuneable electrical components, paving the way for a new generation of bioelectronic devices and sensors. The University of Bristol-led study, published today in The Proceedings of the National Academy of Sciences (PNAS), demonstrates how to make conductive, biodegradable wires from designed proteins. These could be compatible with conventional electronic components made from copper or iron, as well as the biological machinery responsible for generating energy in all living organisms. The miniscule wires are the size of transistors on silicon chips or one thousandth of the breadth of the finest human hair. They are made completely of natural amino acids and heme molecules, found in proteins such as hemoglobin, which transports oxygen in red blood cells. Harmless bacteria were used for their manufacture, eliminating the need for potentially complex and environmentally damaging procedures commonly used in the production of synthetic molecules. Lead author Ross Anderson, Professor of Biological Chemistry at the University of Bristol, said: "While our designs take inspiration from the protein-based electronic circuits necessary for all life on Earth, they are free from much of the complexity and instability that can prevent the exploitation of their natural equivalents on our own terms. We can also build these minute electronic components to order, specifying their properties in a way that is not possible with natural proteins." Leading experts in biomolecular engineering and simulation worked together to produce this unique new method of designing tailor-made proteins with tuneable electronic properties. The multidisciplinary team used advanced computational tools to design simple building blocks that could be combined into longer, wire-like protein chains for conducting electrons. They were able to visualise the structures of these wires using protein X-ray crystallography and electron cryo-microscopy (cryo-EM), techniques which allow structures to be viewed in the finest detail. Pushing the technical boundaries of cryo-EM, images of the smallest individual protein ever studied were obtained with this technique. Ultimately, these nanoscale designer wires have the potential to be used in a wide range of applications, including biosensors for the diagnosis of diseases and detection of environmental pollutants. It is also hoped this invention will form the foundation of new electrical circuits for creating tailor-made catalysts for green industrial biotechnology and artificial photosynthetic proteins for capturing solar energy. The breakthrough was possible thanks to a £4.9 million grant from the Biotechnology and Biological Science Research Council (BBSRC), the UK's largest bioscience funder, which resulted in a five-year project entitled 'The Circuits of Life' involving the Universities of Bristol, Portsmouth, East Anglia, and University College London (UCL). The team harnessed their expertise in protein design, electron transfer, biomolecular simulation, structural biology and spectroscopy, gaining insight into how electrons flow through natural biological molecules, a fundamental process which underpins cellular respiration and photosynthesis. Further advances are expected as the project, which began last year, progresses, presenting significant opportunities to help meet the transition to net zero and more sustainable industrial processes. Co-author Adrian Mulholland, Professor of Chemistry at the University of Bristol, said: "These proteins show how protein design is increasingly delivering practically useful tools. They offer exciting possibilities as components for engineering biology and also are great systems for investigating the fundamental mechanisms of biological electron transfer." Paper 'An expandable, modular de novo protein platform for precision redox engineering' by George H. Hutchins, Claire E.M. Noble, Adrian Bunzel et al published in PNAS |
| Type Of Material | Technology assay or reagent |
| Year Produced | 2023 |
| Provided To Others? | Yes |
| Impact | Pioneering study signals new era of environment-friendly programmable bioelectronics Image shows the design of a protein nanowire, with the green arrow indicating electron flow. Ross Anderson Image shows structural analysis of the protein-based wire, comparing the model of the designed protein (shown in red) with the experimentally determined structure (in grey). Ross Anderson Press release issued: 25 July 2023 Researchers have created a unique microscopic toolkit of 'green' tuneable electrical components, paving the way for a new generation of bioelectronic devices and sensors. The University of Bristol-led study, published today in The Proceedings of the National Academy of Sciences (PNAS), demonstrates how to make conductive, biodegradable wires from designed proteins. These could be compatible with conventional electronic components made from copper or iron, as well as the biological machinery responsible for generating energy in all living organisms. The miniscule wires are the size of transistors on silicon chips or one thousandth of the breadth of the finest human hair. They are made completely of natural amino acids and heme molecules, found in proteins such as hemoglobin, which transports oxygen in red blood cells. Harmless bacteria were used for their manufacture, eliminating the need for potentially complex and environmentally damaging procedures commonly used in the production of synthetic molecules. Lead author Ross Anderson, Professor of Biological Chemistry at the University of Bristol, said: "While our designs take inspiration from the protein-based electronic circuits necessary for all life on Earth, they are free from much of the complexity and instability that can prevent the exploitation of their natural equivalents on our own terms. We can also build these minute electronic components to order, specifying their properties in a way that is not possible with natural proteins." Leading experts in biomolecular engineering and simulation worked together to produce this unique new method of designing tailor-made proteins with tuneable electronic properties. The multidisciplinary team used advanced computational tools to design simple building blocks that could be combined into longer, wire-like protein chains for conducting electrons. They were able to visualise the structures of these wires using protein X-ray crystallography and electron cryo-microscopy (cryo-EM), techniques which allow structures to be viewed in the finest detail. Pushing the technical boundaries of cryo-EM, images of the smallest individual protein ever studied were obtained with this technique. Ultimately, these nanoscale designer wires have the potential to be used in a wide range of applications, including biosensors for the diagnosis of diseases and detection of environmental pollutants. It is also hoped this invention will form the foundation of new electrical circuits for creating tailor-made catalysts for green industrial biotechnology and artificial photosynthetic proteins for capturing solar energy. The breakthrough was possible thanks to a £4.9 million grant from the Biotechnology and Biological Science Research Council (BBSRC), the UK's largest bioscience funder, which resulted in a five-year project entitled 'The Circuits of Life' involving the Universities of Bristol, Portsmouth, East Anglia, and University College London (UCL). The team harnessed their expertise in protein design, electron transfer, biomolecular simulation, structural biology and spectroscopy, gaining insight into how electrons flow through natural biological molecules, a fundamental process which underpins cellular respiration and photosynthesis. Further advances are expected as the project, which began last year, progresses, presenting significant opportunities to help meet the transition to net zero and more sustainable industrial processes. Co-author Adrian Mulholland, Professor of Chemistry at the University of Bristol, said: "These proteins show how protein design is increasingly delivering practically useful tools. They offer exciting possibilities as components for engineering biology and also are great systems for investigating the fundamental mechanisms of biological electron transfer." Paper 'An expandable, modular de novo protein platform for precision redox engineering' by George H. Hutchins, Claire E.M. Noble, Adrian Bunzel et al published in PNAS |
| URL | https://www.bristol.ac.uk/news/2023/july/protein-nanowires.html |
| Title | Annual report to EPSRC; details below |
| Description | HECBioSim, the UK HEC Biomolecular Simulation Consortium, was established in March 2013, and works closely with and complements CCP-BioSim (the UK Collaborative Computational Project for Biomolecular Simulation at the Life Sciences Interface). Most of the members of the Consortium are experienced users of high end computing. Biomolecular simulations are now making significant contributions to a wide variety of problems in drug design and development, biocatalysis, bio and nano-technology, chemical biology and medicine. The UK has a strong and growing community in this field, recognized by the establishment in 2011 by the EPSRC of CCP-BioSim (ccpbiosim.ac.uk), and its renewal in 2015. There is a clear, growing and demonstrable need for HEC in this field. Members of the Consortium have, for example, served on the HECToR and ARCHER Resource Allocation Panel. The Consortium welcomes members across the whole biomolecular sciences community, and we are currently (and will remain) open to new members (unlike some consortia). Since establishing the Consortium, several new members (FLG, DH, ER) have joined the Management Group. Many of the projects awarded ARCHER time under the Consortium do not involve CCP-BioSim or HECBioSim Management Group members, demonstrating the openness of HECBioSim and its support of the biomolecular simulation community in the UK. A number of other researchers have joined and it is our expectation that other researchers will join HECBioSim in the future. We actively engage with structural and chemical biologists and industrial researchers. We foster interactions between computational, experimental and industrial scientists (see the example case studies; members of the Consortium have excellent links with many pharmaceutical, chemical and biotechnology companies). Details of HECBioSim are available via our webpages at: http://www.hecbiosim.ac.uk Applications are made through the website http://www.hecbiosim.ac.uk and reviewed at one of the series of regular panel meetings. A list of successful HECtime allocations on ARCHER is available at:http://www.hecbiosim.ac.uk/applications/successfulprojects All proposals are of course subject to scientific and technical review, but we have the philosophy of supporting the best science to deliver the highest impact, rather than focusing on supporting development of a couple of codes. An allocation panel (with changing membership) meets twice yearly to judge proposals and requests for AUs; projects are assessed competitively: any groups in the UK can apply. All submitted proposals receive constructive feedback from the allocation panel. The HECBioSim website also provides forums for the biomolecular simulation community, a wiki hosting useful how-tos and user guides, and software downloads (currently FESetup and Longbow). Our lead software development project between Nottingham (Charlie Laughton and Gareth Shannon) and Daresbury (James Gebbie), we have developed a remote job submission tool, 'Longbow' (see below). The tool is designed to reproduce the look and feel of local MD packages, but to stage data and submit jobs to a large HPC resource, such as ARCHER, in a manner invisible to the user. Workshops and New Opportunities No more than half a page detailing upcoming workshops and meetings. Also include discussion of other areas which are potentially of interest to other HEC Consortia and opportunities for cross CCP /HEC working. Workshops are listed below and can be found on the HECBioSim Website We work closely with CCP-BioSim, for example in organizing training workshops and meetings. Our activities are outlined at www.hecbiosim.ac.uk Examples of forthcoming meetings include: Computational Molecular Science 2017 - on Sunday 19 March 2017 17th European Seminar on Computational Methods in Quantum Chemistry - on Tuesday 11 July 2017 Frontiers of Biomolecular Simulation Southampton, September 2016 Issues and Problems Short discussion of any issues or problems that the HEC Consortium faces including issues with the ARCHER service, funding, management of allocation and staffing. This is also a good opportunity to highlight to the SLA committee any issues that the Consortium may have experienced with their SLA support during the reporting period. Details of SLA activities are given in the SLA report. Dr. James Gebbie provides support to HECBioSIm through the SLA. He has been very helpful in the construction of the HECBioSim webpages (hecbiosim.ac.uk). James also worked with Dr. Gareth Shannon on the development of Longbow (See above). In the past year, members of the Consortium faced some queuing problems; close to the end of the first allocation period in particular, there were long queuing times, which led to problems in completing jobs and using allocated time. Throughput has been a real problem in the past few months. Membership Please provide a full list of existing members and their institutions, highlighting any new members that have joined the consortium during the reporting period. If available please provide information on the number of distinct users that have accessed ARCHER via the Consortium during this reporting period. Below is a list of PIs with HECBioSim projects in the current reporting period: Agnes Noy, University of York - new to this reporting period Alessandro Pandini, Brunel University London Arianna Fornili, Queen Mary University of London Cait MacPhee, University of Edinbrurgh - new to this reporting period Charlie Laughton, University of Nottingham Clare-Louise Towse, University of Bradford - new to this reporting period D Flemming Hansen, University College London - new to this reporting period David Huggins, Aston University Edina Rosta, King's College London Francesco Gervasio, University College London Ian Collinson, University of Bristol - new to this reporting period Irina Tikhonova, Queen's University Belfast Jiayun Pang, University of Greenwich Jonathan Doye, University of Oxford Julien Michel, University of Edinburgh Mario Orsi, UWE Bristol Michele Vendruscolo, University of Cambridge Michelle Sahai, University of Roehampton Philip Biggin, University of Oxford Richard Sessions, University of Bristol Stephen Euston, Heriot-Watt University Syma Khalid, University of Southampton PIs in HECBioSIm outside of current allocation period Peter Bond, University of Cambridge Richard Bryce, University of Manchester Juan Antonio Bueren-Calabuig, University of Edinburgh Christo Christov, Northumbria University Anna K Croft, University of Nottingham Jonathan Essex, University of Southampton Robert Glen, University of Cambridge Sarah A Harris, University of Leeds Jonathan Hirst, University of Nottingham Douglas Houston, University of Edinburgh Dmitry Nerukh, Aston University Andrei Pisliakov, University of Dundee Mark Sansom, University of Oxford Marieke Schor, University of Edinburgh Gareth Shannon, University of Nottingham Tatyana Karabencheva-Christova, Northumbria University There are currently 164 members of HECBioSim (i.e. users of HECBioSIm ARCHER time), included the above PIs. 20 new users have been added since January 2016. World class and world leading scientific output: ARCHER should enable high quality and world-leading science to be delivered. This should generate high impact outputs and outcomes that increase the UK's position in world science. • If all the publications relating to the work of the Consortium for this reporting period have been added to ResearchFish / will be added to ResearchFish by the end of the ResearchFish reporting exercise, please indicate this below. • If submission of a full list of publications to the Consortium record/s in ResearchFish has not been possible for this reporting period please provide a list of publications that have resulted from work performed on ARCHER by the Consortium during this reporting period (this can be included as a separate attachment). • For the reporting period please provide a bullet pointed list of key / important research findings that has resulted from work performed on ARCHER by the Consortium. Please reference any related publications. • For the reporting period please include a bullet pointed list of any relevant press announcements and other communications of significance to an international community. All publications can be found in ResearchFish, and have been added by individual researchers. A selection of illustrative highlights are provided below: Julien Michel (University of Edinburgh) New molecular simulation methods have enabled for the first time a complete description of the interactions of several drug-like small molecules with an intrinsically disordered region of the oncoprotein MDM2, as well as the effect of post-translational modifications on the dynamics of this protein. The results obtained suggest new medicinal chemistry strategies to achieve potent and selective inhibition of MDM2 for cancer therapies. Publication: Elucidation of Ligand-Dependent Modulation of Disorder-Order Transitions in the Oncoprotein MDM2 http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1004282 Impact of Ser17 Phosphorylation on the Conformational Dynamics of the Oncoprotein MDM2 http://pubs.acs.org/doi/abs/10.1021/acs.biochem.6b00127 An outreach document is being prepared by EPCC to showcase a lab project on isoform selective inhibition that has benefited from both HECBioSim and EPCC ARCHER allocations. Syma Khalid (Chemistry, Southampton) Publication: OmpA: A Flexible Clamp for Bacterial Cell Wall Attachment Samsudin F, Ortiz-Suarez ML, Piggot TJ, Bond PJ, Khalid S (2016). "OmpA: a Flexible Clamp for Bacterial Cell Wall Attachment", Structure, 24(12):2227-2235 With Singapore National Center for Biotechnology. Sarah Harris (Physics, Leeds) In collaboration with the experimental biochemistry group lead by Radford at Leeds, Sarah Harris used ARCHER time to show how chaperones help outer membrane proteins to fold, see: Schiffrin B, Calabrese AN, Devine PWA, Harris SA, Ashcroft AE, Brockwell DJ, Radford SE "Skp is a multivalent chaperone of outer-membrane proteins" Nature Structural and Molecular Biology 23 786-793, 2016. DOI:10.1038/nsmb.3266 Michelle Sahai (Department of Life Sciences, Roehampton) Publication: Combined in vitro and in silico approaches to the assessment of stimulant properties of novel psychoactive substances - The case of the benzofuran 5-MAPB. Progress in Neuro-Psychopharmacology and Biological Psychiatry 75, Pages 1-9 (2017) https://www.ncbi.nlm.nih.gov/pubmed/27890676 Phil Biggin (Biochemistry, Oxford) Outputs that have used HECBioSim time: 1. Steered Molecular Dynamics Simulations Predict Conformational Stability of Glutamate Receptors. Musgaard M, Biggin PC. J Chem Inf Model. 2016 Sep 26;56(9):1787-97. doi: 10.1021/acs.jcim.6b00297. 2. Kainate receptor pore-forming and auxiliary subunits regulate channel block by a novel mechanism. Brown PM, Aurousseau MR, Musgaard M, Biggin PC, Bowie D. J Physiol. 2016 Apr 1;594(7):1821-40. doi: 10.1113/JP271690. 3. Role of an Absolutely Conserved Tryptophan Pair in the Extracellular Domain of Cys-Loop Receptors. Braun N, Lynagh T, Yu R, Biggin PC, Pless SA. ACS Chem Neurosci. 2016 Mar 16;7(3):339-48. doi: 10.1021/acschemneuro.5b00298. 4. Distinct Structural Pathways Coordinate the Activation of AMPA Receptor-Auxiliary Subunit Complexes. Dawe GB, Musgaard M, Aurousseau MR, Nayeem N, Green T, Biggin PC, Bowie D. Neuron. 2016 Mar 16;89(6):1264-76. doi: 10.1016/j.neuron.2016.01.038. Covered by 7 news outlets: I. Health Canal (http://www.healthcanal.com/brain-nerves/70646-what-makes-the-brain-tick-so-fast.html) II. Health Medicine Network (http://healthmedicinet.com/i/scientists-reveal-how-the-brain-processes-information-with-lightning-speed/) III. News Medical (http://www.news-medical.net/news/20160227/Scientists-reveal-how-the-brain-processes-information-with-lightning-speed.aspx) IV. Wired (http://www.wired.co.uk/article/what-we-learned-about-the-brain-this-week-3) V. Alpha Galileo (http://www.alphagalileo.org/ViewItem.aspx?ItemId=161473&CultureCode=en) VI. Science Daily (https://www.sciencedaily.com/releases/2016/02/160225140254.htm) VII. Eureka Alert (https://www.eurekalert.org/pub_releases/2016-02/mu-wmt022516.php) Recommended by F1000 Prime: http://f1000.com/prime/726180600. 5. Accurate calculation of the absolute free energy of binding for drug molecules. Aldeghi M, Heifetz A, Bodkin MJ, Knapp S, Biggin PC. Chem Sci. 2016 Jan 14;7(1):207-218. Arianna Fornili (Biological and Chemical Sciences, Queen Mary University of London) Publication: Fornili, E. Rostkova, F. Fraternali, M. Pfuhl: Effect of RlC N-Terminal Tails on the Structure and Dynamics of Cardiac Myosin. Biophysical J. 110 (2016) 297A. More in preparation. Dmitry Nerukh (Engineering and Applied Science, Aston University) Important research findings: Distribution of ions inside a viral capsids influences the stability of the capsid Edina Rosta (Computational Chemistry, Kings College London) Highlighted publication: We have a joint experimental paper with the Vertessy group in JACS where we identified a novel arginine finger residue in dUTPase enzymes, and described the mechanism of action for this residue using crystallographic data, biochemical experiments, MD and QM/MM simulations thanks to HECBioSim. http://pubs.acs.org/doi/abs/10.1021/jacs.6b09012 Press release: Our joint paper with Jeremy Baumberg's group in Cambridge was published in Nature. http://www3.imperial.ac.uk/newsandeventspggrp/imperialcollege/newssummary/news_13-6-2016-16-52-4 Greater scientific productivity: As well as speed increases, the optimisation of codes for the ARCHER machine will enable problems to be solved in less time using fewer compute resources. • For the reporting period please provide a brief update on the progress of software development activities associated with the Consortium and the impact this has had on Consortium members and the broader research community. Our lead software development project between Nottingham (Charlie Laughton and Gareth Shannon) and Daresbury (James Gebbie), we have developed a remote job submission tool, 'Longbow' (see below). The tool is designed to reproduce the look and feel of local MD packages, but to stage data and submit jobs to a large HPC resource such as ARCHER in a manner invisible to the user. An open beta version was released unrestricted to the community in March 2015 (http://www.hecbiosim.ac.uk/longbow and via PyPI https://pypi.python.org ) with ~100 downloads. Functionality includes native support for biosimulation packages AMBER, CHARMM, GROMACS, LAMMPS and NAMD, native support for jobs on ARCHER, native support for jobs running on PBS and LSF schedulers, and support for three different job types (single, ensembles and multiple jobs). The software is written both as an application for users and an API for developers. CCP-EM have integrated Longbow into their developmental GUI, and shortly FESetup will ship with native support for job submission using Longbow. We are currently in talks with ClusterVision with respect to them distributing Longbow to their user base as a user friendly way for novices to interact with their systems. Longbow Longbow has been growing in popularity both amongst researchers within the biosimulation field and those in other fields. In this reporting period there have been five new releases of Longbow delivering a plethora of user requested features and bug fixes. Some of the key developments are summarised below: • Implemented a Recovery mode - Should the event happen that Longbow crashes or the system in which Longbow is controlling jobs from powers down. The user can now reconnect to the crashed session and carry on as if nothing happened. • Sub-Queuing - More and more system administrators are setting limits not just on the number of simulations that can run, but also on the number of jobs that can go into the queue. Longbow can now automatically detect this and implement its own queue feeding jobs into the system queue as slots open up. • Dis-connectible/Re-connectible Longbow sessions - A user can now launch Longbow to fire off all jobs and then disconnect, at a later date the user can re-establish the connection and download all results (no need for persistent connections anymore) • Ability to include scripts in the Longbow generated submit files. • Numerous stability, performance and bug fixes Longbow continues to be the submission engine that is used under the hood of the CCP-EM toolkit FLEX-EM. There are several other interesting projects that are making use of Longbow. A project by the energy efficient computing group (Hartree Center) are currently integrating Longbow with CK, a crowd sourced compilation optimisation tool aimed at finding efficient compilation configuration. A project by the Applications performance engineering group (Hartree Center) are currently integrating longbow into Melody, a runtime optimisation tool aimed at optimising runtime configuration of simulations. A project by the computational biology group (Hartree Center) an automated setup, launch and analysis platform for MD on protein-membrane simulations. Metrics Downloads (HECBioSim) 985 Downloads (PyPi) 3456 Increasing the UK's CSE skills base (including graduate and post doctorate training and support): This builds on the skills sets of trained people in HPC, both in terms of capacity and raising the overall skill level available to the sector. • For the reporting period please provide information on the number of PhDs and Post-Docs that have been trained in the use of ARCHER as a result of work relating to the Consortium. • For the reporting period please provide a bullet pointed list of training activities undertaken by the Consortium, providing information on the target audience and level of attendance. 125 PhD and PDRAs have been trained and have used ARCHER through HECBioSim. In the past 6 months alone that has included: PhD students: Marc Dämgen, University of Oxford Laura Domicevica, University of Oxford Matteo Aldeghi, University of Oxford Shaima Hashem, Queen Mary University of London Ruth Dingle, University College London Lucas Siemons, University College London Elvira Tarasova, Aston University Vladimiras Oleinikovas, University College London Havva Yalinca, University College London Daniel Moore, Queen's University Belfast Aaron Maguire, Queen's University Belfast Maxime Tortora, University of Oxford Michail Palaiokostas, UWE Bristol Wei Ding, UWE Bristol Ganesh Shahane, UWE Bristol Pin-Chia Hsu, University of Southampton Damien Jefferies, University of Southampton PDRA: Maria Musgaard, University of Oxford Teresa Paramo, University of Oxford Marieke Schor, University of Edinbrurgh Antonia Mey, University of Edinbrurgh Ioanna Styliari, University of Nottingham Ivan Korotkin, Aston University Silvia Gomez Coca, King's College London Giorgio Saladino, University College London Federico Comitani, University College London Robin Corey, University of Bristol Jordi Juarez-Jimenez, University of Edinburgh Predrag Kukic, University of Cambridge Giulia Tomba, University of Cambridge Deborah Shoemark, University of Bristol Georgios Dalkas, Heriot-Watt University Robin Westacott, Heriot-Watt University Firdaus Samsudin, University of Southampton Agnes Noy, University of Leeds (now EPSRC fellow at York). Please see below details of the training week held in June 2016 at the University of Bristol. 170 delegates attending the training which ran over 5 days. There is course content available on the HECBioSim Workshop Page CCPBioSim Training Week - Day 5: QM/MM enzyme reaction modelling - on Friday 10 June 2016 CCPBioSim Training Week - Day 4: Monte Carlo Methods for Biomodelling - on Thursday 09 June 2016 CCPBioSim Training Week - Day 3: Python for Biomodellers and FESetup - on Wednesday 08 June 2016 CCPBioSim Training Week - Day 2: Running and analysing MD simulations - on Tuesday 07 June 2016 CCPBioSim Training Week - Day 1: Enlighten: Tools for enzyme-ligand modelling - on Monday 06 June 2016 Past Workshops and Conferences: AMOEBA advanced potential energies workshop - on Friday 09 December 2016 Intel Training Workshop (Parallel programming and optimisation for Intel architecture) - on Wednesday 30 November 2016 3rd Workshop on High-Throughput Molecular Dynamics (HTMD) 2016 - on Thursday 10 November 2016 Free Energy Calculation and Molecular Kinetics Workshop - on Tuesday 13 September 2016 Going Large: tools to simplify running and analysing large-scale MD simulations on HPC resources - HECBioSim - on Wednesday 16 December 2015 This 1 day workshop dealt with Longbow, a Python tool created by HECBioSim consortium that allows use of molecular dynamics packages (AMBER, GROMACS, LAMMPS, NAMD) with ease from the comfort of the desktop, and pyPcazip, a flexible Python-based package for the analysis of large molecular dynamics trajectory data sets. Richard Henchman ( Chemistry, Manchester) I lectured on the CCP5 Summer School in 2015 in Manchester and 2016 in Lancaster for the Advanced Topic Biomolecular Simulation drawing on CCPBiosim training materials. Sarah Harris (Physics, Leeds) Agnes Noy Agnes Noy was trained to use ARCHER independently which helped her to obtain an Early-career EPSRC fellowship. The fellowship grant follows the study of supercoiling DNA by modelling (further funding) and introduces a new member to the community (grant is EPSRC (EP/N027639/1)). Increased impact and collaboration with industry: ARCHER does not operate in isolation and the 'impact' of ARCHER's science is converted to economic growth through the interfaces with business and industry. In order to capture the impacts, which may be economic, social, environmental, scientific or political, various metrics may be utilised. • For the reporting period please provide where possible information on Consortium projects that have been performed in collaboration with industry, this should include: o Details of the companies involved. o Information on the part ARCHER and the Consortium played. o A statement on the impact that the work has / is making. o If relevant, details of any in kind or cash contributions that have been associated with this work. • For the reporting period include a list of Consortium publications that have industrial co-authorship. • For the reporting period please provide details of the any other activities involving industrial participation e.g. activities involving any Industrial Advisory panels, attendance / participation in workshops and Consortium based activities. Examples of industrial engagement through HECBioSim include: Syma Khalid (Computational Biophysics, Southampton) Collaboration with Siewert-Jan Marrink (Netherlands) and Wonpil Im(USA) - paper is being written at the moment - also we have developed and tested the following computational tool as part of this project: http://www.charmm-gui.org/?doc=input/membrane We have signed an NDA with a company developing antibiotics (Auspherix) based on work published in Samsudin et al, Structure, 2016. The company provide us with experimental data and crucially the structures of their potential drug molecules, and we work out how they interact with the bacterial membrane. Press releases https://www.sciencedaily.com/releases/2016/11/161121094122.htm ARCHER time we got from HECBioSim was essential for this project. These are large simulations that are very computationally demanding. Julien Michel (Chemistry, Edinburgh) Industrial collaborations that have benefited from HECBioSim UCB was a project partner of EPSRC-funded research in the Michel lab on modelling binding of ligands to flexible proteins. HECBioSim supported the research via allocation of computing time on ARCHER. The work has been published. Impact of Ser17 phosphorylation on the conformational dynamics of the oncoprotein MDM2 Bueren-Calabuig, J. A. ; Michel, J. Biochemistry, 55 (17), 2500-2509, 2016 Elucidation of Ligand-Dependent Modulation of Disorder-Order Transitions in the Oncoprotein MDM2 Bueren-Calabuig, J. A. ; Michel, J. PLoS Comput. Biol. , 11(6): e1004282, 2015 International activities I represented HECBioSim at the MolSSI conceptualisation workshop in Houston (October 2016). Arianna Fornili (School of Biological and Chemical Sciences, Queen Mary University of London) This project is done in collaboration with experimental (NMR and SAXS) partners at King's College London (Dr. Mark Pfuhl and Dr. Elena Rostkova), with the final aim of providing a complete molecular model of how muscle contraction is regulated in the heart. Francesco Gervasio (Chemistry, UCL) A new collaboration with UCB on developing new approaches to sample cryptic binding sites of pharmaceutical interest was started. UCB co-sponsored a BBSRC-CASE PhD studentships. The code development and its application to interesting targets was made possible by the access to Archer. Phil Biggin (Biochemistry, Oxford) The following has industrial partners as co-authors: Accurate calculation of the absolute free energy of binding for drug molecules, Aldeghi M, Heifetz A, Bodkin MJ, Knapp S, Biggin PC. Chem. Sci., 2016,7, 207-218 doi: 10.1039/C5SC02678D. The GLAS scoring module for gromacs has yet to be implemented but that would also constitute industrial activity. Richard Henchman (Chemistry, Manchester) Industrial Collaboration I have one publication with an industrial co-author (Evotec), which Julien Michel is part of as well. Assessment of hydration thermodynamics at protein interfaces with grid cell theory, G. Gerogiokas, M. W. Y. Southey, M. P. Mazanetz, A. Heifetz, M. Bodkin, R. J. Law, R. H. Henchman and J. Michel, J. Phys. Chem. B, 2016, 120, 10442-10452. Strengthening of UK's international position: The impacts of ARCHER's science extend beyond national borders and most science is delivered through partnerships on a national or international level. • For the reporting period please provide a bullet pointed list of projects that have involved international collaboration. For each example please provide a brief summary of the part that ARCHER and the Consortium have played. • For the reporting period please provide a list of consortium publications with international co-authorship. • For the reporting period please detail any other international activities that the Consortium might be involved in (workshops, EU projects etc.). Examples of HECBioSim international engagement include: Michelle Sahai (Department of Life Sciences, Roehampton) Combined in vitro and in silico approaches to the assessment of stimulant properties of novel psychoactive substances - The case of the benzofuran 5-MAPB. Sahai MA, Davidson C, Khelashvili G, Barrese V, Dutta N, Weinstein H, Opacka-Juffry J. Prog Neuropsychopharmacol Biol Psychiatry. (2016) 75:1-9. doi: 10.1016/j.pnpbp.2016.11.004. co-authors affiliations include: Department of Physiology and Biophysics, Weill Cornell Medical College of Cornell University (WCMC), New York, NY, 10065, USA and HRH Prince Alwaleed Bin Talal Bin Abdulaziz Alsaud Institute of Computational Biomedicine, Weill Cornell Medical College of Cornell University, New York, NY, 10065, USA. Author contribution: CD, MAS, HW and JOJ were responsible for the study concept and design. CD and VB collected and interpreted the voltammetry data. JOJ and ND conducted the ligand binding experiments and ND analysed the data. MAS (ARCHER user) and GK performed the molecular modelling studies and interpreted the findings. CD, MAS, GK and JOJ drafted the manuscript. All authors critically reviewed the content and approved the final version for publication. Dmitry Nerukh (Engineering and Applied Science, Aston University) Collaborations: • Prof. Makoto Taiji group (K-computer, MDGRAPE), Laboratory for Computational Molecular Design, Computational Biology Research Core, RIKEN Quantitative Biology Center (QBiC), Kobe, Japan • Prof. Reza Khayat group, City College of New York, United States • Prof. Artyom Yurov group, Immanuel Kant Baltic Federal University, Russian Federation • Prof. Nikolay Mchedlov-Petrosyan group, V.N. Karazin Kharkiv National University, Ukraine • Prof. Natalya Vaysfeld group, Odessa I.I.Mechnikov National University, Ukraine International activities: • International Workshop - Engineering bacteriophages for treating antimicrobial resistance using computational models. Dec. 2016, Aston University, UK Dmitry Nerukh group at Aston University collaborates with the group of Prof. Makoto Taiji (the deputy director of RIKEN Quantitative Biology Institute). Prof. Taiji group is the author and implementer of the fastest machine in the world for molecular dynamics simulations, MDGRAPE, as well as part of the team designing the K-computer and working on it (several times the fastest computer in the world). We are currently preparing a joint manuscript for publication where the results obtained on ARCHER will be used alongside with the results obtained on MDGRAPE-4. Julien Michel (Chemistry, Edinburgh) Free Energy Reproducibility Project Free energy reproducibility project between the Michel, Mobley, Roux groups and STFC. This makes use of FESetup which is CCPBioSim, but calculations have been executed on various HPC platforms, Hannes can clarify whether any of the systems used is within the remit of HECBioSim. Sarah Harris (Physics, Leeds), Charlie Laughton (Pharmacy, Nottingham) and Agnes Noy (Physics, York) In an international collaboration with the Institute for Research in Biomedicine in Barcelona, Noy, Harris and Laughton used ARCHER time obtained through the HecBioSim consortium to perform multiple 100ns molecular dynamics (MD) simulations of DNA minicircles containing ~100 base pairs as part of the validation suite for the new BSC1 nucleic acid forcefield, see: Ivani I., Dans P. D., Noy A., Perez A., Faustino I., Hospital A., Walther J., Andrio P., Goni R., Portella G., Battistini F.,Gelpi J. L. Gonzalez C., Vendruscolo M., Laughton C. A., Harris S. A. Case D. A. & Orozco M. "Parmbsc1: a refined force field for DNA simulations" Nat. Methods 13, 55, 2015, doi:10.1038/nmeth.3658 Jon Essex (Chemistry, Southampton) Development and testing of hybrid coarse-grain/atomistic model of membrane systems. ARCHER and the consortium were instrumental in providing the computing time necessary to complete this work. The work was performed by a research fellow in my group, who has subsequently taken up an academic post in Sweden, where he is continuing to develop this methodology. Publication: All-atom/coarse-grained hybrid predictions of distributioncoefficients in SAMPL5 Samuel Genheden1, Jonathan W. Essex, J Comput Aided Mol Des (2016) 30:969-976 DOI 10.1007/s10822-016-9926-z IP and other industrial engagement, or translation, which has benefited from HECBioSim international collaborations Collaboration Dr Samuel Genheden, University of Gothenburg - see above International activities (e.g. workshops) that have benefited from HECBioSim The work performed with Sam resulted in an successful eCSE application for development work on the LAMMPS software. This is an international simulation package run out of Sandia labs in the US. Through this eCSE application we were able to update the rotational integrator and improve the load balancing for our hybrid simulation methodology. These developments have been incorporated in the latest release versions of the code. An ARCHER training webinar resulted from this work. Training activities see above - ARCHER training webinar on our developments within LAMMPS Software development see above - code modifications in LAMMPS which either have been, or will be, in the release version of the code Advantages you see of the consortium (e.g. could be ability to pursue project quickly; develop collaborations) resources to run the sort of large-scale calculations we need Michelle Sahai (Department of life Sciences, Roehampton) The publication I added on ResearchFish has international co-authors. Combined in vitro and in silico approaches to the assessment of stimulant properties of novel psychoactive substances - The case of the benzofuran 5-MAPB. Sahai MA, Davidson C, Khelashvili G, Barrese V, Dutta N, Weinstein H, Opacka-Juffry J. Prog Neuropsychopharmacol Biol Psychiatry. (2016) 75:1-9. doi: 10.1016/j.pnpbp.2016.11.004. co-authors affiliations include: Department of Physiology and Biophysics, Weill Cornell Medical College of Cornell University (WCMC), New York, NY, 10065, USA and HRH Prince Alwaleed Bin Talal Bin Abdulaziz Alsaud Institute of Computational Biomedicine, Weill Cornell Medical College of Cornell University, New York, NY, 10065, USA. Author contribution: CD, MAS, HW and JOJ were responsible for the study concept and design. CD and VB collected and interpreted the voltammetry data. JOJ and ND conducted the ligand binding experiments and ND analysed the data. MAS (ARCHER user) and GK performed the molecular modelling studies and interpreted the findings. CD, MAS, GK and JOJ drafted the manuscript. All authors critically reviewed the content and approved the final version for publication. Edina Rosta (Computational Chemistry, Kings College London) International collaborations I have several international collaborators with joint computational/experimental projects. For their success the HECBioSim computer time was essential: • Prof. Beata Vertessy, Budapest Technical University, Hungary JACS 2016 138 (45), 15035-15045 • Profs. Walter Kolch, Vio Buchete and Boris Kholodenko, UCD, Ireland Angewandte 55 (3), 983-986, 2016 (this may have been reported previously) PLOS Computational Biology 12 (10), e1005051, 2016 J Phys Chem Letters 7 (14), 2676-2682, 2016 • Jose Maria Lluch, Angels Gonzalez, Autonomous University of Barcelona, Spain JCTC 12 (4), 2079-2090, 2016 Other international activities Free energy and molecular kinetics workshop with Frank Noe's group (Free University of Berlin) and Vio Buchete (UCD, Dublin). Maybe this should be for training activities? Following the workshop, Erasmus students apply to visit my group. Phil Biggin (Biochemistry, Oxford) 2 and 4 were with colleagues from McGill (Brown PM, Aurousseau MR and Bowie) and 3 was with colleagues from Denmark (Braun and Pless) 2. Kainate receptor pore-forming and auxiliary subunits regulate channel block by a novel mechanism. Brown PM, Aurousseau MR, Musgaard M, Biggin PC, Bowie D. J Physiol. 2016 Apr 1;594(7):1821-40. doi: 10.1113/JP271690. 3. Role of an Absolutely Conserved Tryptophan Pair in the Extracellular Domain of Cys-Loop Receptors. Braun N, Lynagh T, Yu R, Biggin PC, Pless SA. ACS Chem Neurosci. 2016 Mar 16;7(3):339-48. doi: 10.1021/acschemneuro.5b00298. 4. Distinct Structural Pathways Coordinate the Activation of AMPA Receptor-Auxiliary Subunit Complexes. Dawe GB, Musgaard M, Aurousseau MR, Nayeem N, Green T, Biggin PC, Bowie D. Neuron. 2016 Mar 16;89(6):1264-76. doi: 10.1016/j.neuron.2016.01.038. Richard Henchman I have an international collaboration with Professor Franke Grater at the University of Heidelberg on entropy theory for biomolecular systems. Syma Khalid (Chemistry, Southampton) Collaboration with Singapore National Center for Biotechnology. Contributions to CECAM workshops Other Highlights for the Current Reporting Period: Please provide details of any other significant highlights from the reporting period that are not captured elsewhere in the report. Professor Adrian Mullholland organised and chaired the Computational Chemistry, Gordon Research Conference, Girona, Spain 24th - 29th July 2016. The theme of the 2016 Computational Chemistry GRC was "Theory and Simulation Across Scales in Molecular Science". It focused on method development and state-of-the-art applications across computational molecular science, and encouraged cross-fertilization between areas. The conference was oversubscribed, with attendees from industry and academia. The meeting received a High-Performance Rating, and was commended for the assessment that 95% of conferees rated this meeting "above average" on all evaluation areas (science, discussion, management, atmosphere and suitability). HEC Consortia Model: Over the coming months EPSRC will be looking at the future of the HEC Consortia model and potential future funding. We would like to use this opportunity to ask the Consortia Chairs for input: • What are the key benefits that your community have experienced through the existence of the HEC Consortia? • What elements of the financial support provided by the HEC Consortium's grant have worked well and what could be improved in the future? We aim to expand the breadth of the work of the Consortium focusing on cutting-edge applications, and building collaborations with experiments and industry, to achieve maximum impact from ARCHER use. We discussed Grand Challenges at our most recent Management Group meeting. In December 2015, and have identified several to follow up as strategic priorities. We aim to tackle and support large-scale grand challenge applications in biomolecular science, in areas such as antimicrobial resistance, membrane dynamics, drug design and synthetic biology. One specific theme with potentially high impact in drug discovery is large -scale comparative investigation of allosteric regulation in different superfamilies of proteins (e.g. PAS domain containing proteins, tyrosine kinases, etc.). The major impact will be in the identification of novel selective therapeutic molecules with limited adverse side effects. As a Consortium, we intend to develop and apply novel computational approaches for the rational design of allosteric regulators of hitherto 'undruggable' targets. Many of the targets emerging from large-scale genetic screening are deemed undruggable due to the difficulty of designing drug-like modulators that bind to their catalytic sites. It is increasingly clear that these targets might be effectively targeted by designing drugs that bind to protein-protein interfaces and allosteric sites. To do that, however, new rational design strategies, based on an in-depth knowledge of protein dynamics and advanced modelling and new simulation techniques are required. Other challenging frontier areas include dynamics of motor proteins; prediction of the effects of pathogenic mutations on protein function. The accurate prediction of free energy of binding is still in its infancy and needs much further investigation. Finally, kinetics of biomolecular reactions will become the next big topic. It is clear that high end computing resource can provide the necessary power and capability to provide inroads into this vital area. This is important because it is becoming increasingly apparent that kinetics of drug binding is a key factor that the pharmaceutical sector should really be looking at in terms of a major dictator of potency. HECBioSim is now well established, supporting work of many groups across the UK in the growing field of biomolecular simulation. We benefit from an Advisory Group containing members from industry, as we as international biomolecular simulation experts and experimental scientists. The Advisory Board was expanded and refreshed for the renewal and consists of: Dr. Nicolas Foloppe (Vernalis plc, Chair); Dr. Colin Edge (GlaxoSmithKline); Dr. Mike King (UCB Pharmaceuticals); Dr. Mike Mazanetz (Evotec AG); Dr. Garrett Morris (Crysalin Ltd.); Dr. Gary Tresadern (Johnson and Johnson Pharmaceuticals); Dr. Richard Ward (AstraZeneca); Prof. Modesto Orozco (IRB, Barcelona); Prof. Tony Watts, (Oxford, NMR); Dr. Pete Bond (A*STAR Bioinformatics Institute Singapore). We aim to foster industrial collaborations and collaborations with experimentalists (e.g. joint workshops with CCPN, Institute of Physics (Sarah Harris, Leeds Physics); see e.g. case studies. A particular theme for strategic development will be multiscale modelling, building on collaborations between several groups in the Consortium and the other CCPs. This theme reflects the inclusive and forward looking philosophy of HecBioSim, which is a community open to new ideas, keen to develop new methods, and ultimately to use HEC to drive exciting new science. The Consortium model allows new collaborations to develop. The recent reduction in time allocated to HECBioSim has meant that we can support fewer projects. There is significantly more demand for computer time than we can accommodate through our current allocation. We intend to work with Tier 2 Centres to explore possibilities for applications, and e.g. to test emerging architectures for biomolecular simulation. Edina Rosta (Computational Chemistry, Kings College London) comments: "HECBioSim enables my group to perform biomolecular simulations at the high standards required for JACS, Angewandte, etc. Without this consortium I would not be able to perform the necessary calculations leading to publishable work in top journals as our local university resources are limited and the hpc systems are not well maintained". Regarding financial support: the administrative support provided via the consortium grant is essential to the functioning of HECBioSim. This role is currently undertaken by Simone Breckell who not only administers, allocates time to projects and arranges panel meetings but has also done a fantastic job collating the extensive information for this EPSRC report. |
| Type Of Material | Improvements to research infrastructure |
| Year Produced | 2017 |
| Provided To Others? | Yes |
| Impact | HECBioSim, the UK HEC Biomolecular Simulation Consortium, was established in March 2013, and works closely with and complements CCP-BioSim (the UK Collaborative Computational Project for Biomolecular Simulation at the Life Sciences Interface). Most of the members of the Consortium are experienced users of high end computing. Biomolecular simulations are now making significant contributions to a wide variety of problems in drug design and development, biocatalysis, bio and nano-technology, chemical biology and medicine. The UK has a strong and growing community in this field, recognized by the establishment in 2011 by the EPSRC of CCP-BioSim (ccpbiosim.ac.uk), and its renewal in 2015. There is a clear, growing and demonstrable need for HEC in this field. Members of the Consortium have, for example, served on the HECToR and ARCHER Resource Allocation Panel. The Consortium welcomes members across the whole biomolecular sciences community, and we are currently (and will remain) open to new members (unlike some consortia). Since establishing the Consortium, several new members (FLG, DH, ER) have joined the Management Group. Many of the projects awarded ARCHER time under the Consortium do not involve CCP-BioSim or HECBioSim Management Group members, demonstrating the openness of HECBioSim and its support of the biomolecular simulation community in the UK. A number of other researchers have joined and it is our expectation that other researchers will join HECBioSim in the future. We actively engage with structural and chemical biologists and industrial researchers. We foster interactions between computational, experimental and industrial scientists (see the example case studies; members of the Consortium have excellent links with many pharmaceutical, chemical and biotechnology companies). Details of HECBioSim are available via our webpages at: http://www.hecbiosim.ac.uk Applications are made through the website http://www.hecbiosim.ac.uk and reviewed at one of the series of regular panel meetings. A list of successful HECtime allocations on ARCHER is available at:http://www.hecbiosim.ac.uk/applications/successfulprojects All proposals are of course subject to scientific and technical review, but we have the philosophy of supporting the best science to deliver the highest impact, rather than focusing on supporting development of a couple of codes. An allocation panel (with changing membership) meets twice yearly to judge proposals and requests for AUs; projects are assessed competitively: any groups in the UK can apply. All submitted proposals receive constructive feedback from the allocation panel. The HECBioSim website also provides forums for the biomolecular simulation community, a wiki hosting useful how-tos and user guides, and software downloads (currently FESetup and Longbow). Our lead software development project between Nottingham (Charlie Laughton and Gareth Shannon) and Daresbury (James Gebbie), we have developed a remote job submission tool, 'Longbow' (see below). The tool is designed to reproduce the look and feel of local MD packages, but to stage data and submit jobs to a large HPC resource, such as ARCHER, in a manner invisible to the user. Workshops and New Opportunities No more than half a page detailing upcoming workshops and meetings. Also include discussion of other areas which are potentially of interest to other HEC Consortia and opportunities for cross CCP /HEC working. Workshops are listed below and can be found on the HECBioSim Website We work closely with CCP-BioSim, for example in organizing training workshops and meetings. Our activities are outlined at www.hecbiosim.ac.uk Examples of forthcoming meetings include: Computational Molecular Science 2017 - on Sunday 19 March 2017 17th European Seminar on Computational Methods in Quantum Chemistry - on Tuesday 11 July 2017 Frontiers of Biomolecular Simulation Southampton, September 2016 Issues and Problems Short discussion of any issues or problems that the HEC Consortium faces including issues with the ARCHER service, funding, management of allocation and staffing. This is also a good opportunity to highlight to the SLA committee any issues that the Consortium may have experienced with their SLA support during the reporting period. Details of SLA activities are given in the SLA report. Dr. James Gebbie provides support to HECBioSIm through the SLA. He has been very helpful in the construction of the HECBioSim webpages (hecbiosim.ac.uk). James also worked with Dr. Gareth Shannon on the development of Longbow (See above). In the past year, members of the Consortium faced some queuing problems; close to the end of the first allocation period in particular, there were long queuing times, which led to problems in completing jobs and using allocated time. Throughput has been a real problem in the past few months. Membership Please provide a full list of existing members and their institutions, highlighting any new members that have joined the consortium during the reporting period. If available please provide information on the number of distinct users that have accessed ARCHER via the Consortium during this reporting period. Below is a list of PIs with HECBioSim projects in the current reporting period: Agnes Noy, University of York - new to this reporting period Alessandro Pandini, Brunel University London Arianna Fornili, Queen Mary University of London Cait MacPhee, University of Edinbrurgh - new to this reporting period Charlie Laughton, University of Nottingham Clare-Louise Towse, University of Bradford - new to this reporting period D Flemming Hansen, University College London - new to this reporting period David Huggins, Aston University Edina Rosta, King's College London Francesco Gervasio, University College London Ian Collinson, University of Bristol - new to this reporting period Irina Tikhonova, Queen's University Belfast Jiayun Pang, University of Greenwich Jonathan Doye, University of Oxford Julien Michel, University of Edinburgh Mario Orsi, UWE Bristol Michele Vendruscolo, University of Cambridge Michelle Sahai, University of Roehampton Philip Biggin, University of Oxford Richard Sessions, University of Bristol Stephen Euston, Heriot-Watt University Syma Khalid, University of Southampton PIs in HECBioSIm outside of current allocation period Peter Bond, University of Cambridge Richard Bryce, University of Manchester Juan Antonio Bueren-Calabuig, University of Edinburgh Christo Christov, Northumbria University Anna K Croft, University of Nottingham Jonathan Essex, University of Southampton Robert Glen, University of Cambridge Sarah A Harris, University of Leeds Jonathan Hirst, University of Nottingham Douglas Houston, University of Edinburgh Dmitry Nerukh, Aston University Andrei Pisliakov, University of Dundee Mark Sansom, University of Oxford Marieke Schor, University of Edinburgh Gareth Shannon, University of Nottingham Tatyana Karabencheva-Christova, Northumbria University There are currently 164 members of HECBioSim (i.e. users of HECBioSIm ARCHER time), included the above PIs. 20 new users have been added since January 2016. World class and world leading scientific output: ARCHER should enable high quality and world-leading science to be delivered. This should generate high impact outputs and outcomes that increase the UK's position in world science. • If all the publications relating to the work of the Consortium for this reporting period have been added to ResearchFish / will be added to ResearchFish by the end of the ResearchFish reporting exercise, please indicate this below. • If submission of a full list of publications to the Consortium record/s in ResearchFish has not been possible for this reporting period please provide a list of publications that have resulted from work performed on ARCHER by the Consortium during this reporting period (this can be included as a separate attachment). • For the reporting period please provide a bullet pointed list of key / important research findings that has resulted from work performed on ARCHER by the Consortium. Please reference any related publications. • For the reporting period please include a bullet pointed list of any relevant press announcements and other communications of significance to an international community. All publications can be found in ResearchFish, and have been added by individual researchers. A selection of illustrative highlights are provided below: Julien Michel (University of Edinburgh) New molecular simulation methods have enabled for the first time a complete description of the interactions of several drug-like small molecules with an intrinsically disordered region of the oncoprotein MDM2, as well as the effect of post-translational modifications on the dynamics of this protein. The results obtained suggest new medicinal chemistry strategies to achieve potent and selective inhibition of MDM2 for cancer therapies. Publication: Elucidation of Ligand-Dependent Modulation of Disorder-Order Transitions in the Oncoprotein MDM2 http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1004282 Impact of Ser17 Phosphorylation on the Conformational Dynamics of the Oncoprotein MDM2 http://pubs.acs.org/doi/abs/10.1021/acs.biochem.6b00127 An outreach document is being prepared by EPCC to showcase a lab project on isoform selective inhibition that has benefited from both HECBioSim and EPCC ARCHER allocations. Syma Khalid (Chemistry, Southampton) Publication: OmpA: A Flexible Clamp for Bacterial Cell Wall Attachment Samsudin F, Ortiz-Suarez ML, Piggot TJ, Bond PJ, Khalid S (2016). "OmpA: a Flexible Clamp for Bacterial Cell Wall Attachment", Structure, 24(12):2227-2235 With Singapore National Center for Biotechnology. Sarah Harris (Physics, Leeds) In collaboration with the experimental biochemistry group lead by Radford at Leeds, Sarah Harris used ARCHER time to show how chaperones help outer membrane proteins to fold, see: Schiffrin B, Calabrese AN, Devine PWA, Harris SA, Ashcroft AE, Brockwell DJ, Radford SE "Skp is a multivalent chaperone of outer-membrane proteins" Nature Structural and Molecular Biology 23 786-793, 2016. DOI:10.1038/nsmb.3266 Michelle Sahai (Department of Life Sciences, Roehampton) Publication: Combined in vitro and in silico approaches to the assessment of stimulant properties of novel psychoactive substances - The case of the benzofuran 5-MAPB. Progress in Neuro-Psychopharmacology and Biological Psychiatry 75, Pages 1-9 (2017) https://www.ncbi.nlm.nih.gov/pubmed/27890676 Phil Biggin (Biochemistry, Oxford) Outputs that have used HECBioSim time: 1. Steered Molecular Dynamics Simulations Predict Conformational Stability of Glutamate Receptors. Musgaard M, Biggin PC. J Chem Inf Model. 2016 Sep 26;56(9):1787-97. doi: 10.1021/acs.jcim.6b00297. 2. Kainate receptor pore-forming and auxiliary subunits regulate channel block by a novel mechanism. Brown PM, Aurousseau MR, Musgaard M, Biggin PC, Bowie D. J Physiol. 2016 Apr 1;594(7):1821-40. doi: 10.1113/JP271690. 3. Role of an Absolutely Conserved Tryptophan Pair in the Extracellular Domain of Cys-Loop Receptors. Braun N, Lynagh T, Yu R, Biggin PC, Pless SA. ACS Chem Neurosci. 2016 Mar 16;7(3):339-48. doi: 10.1021/acschemneuro.5b00298. 4. Distinct Structural Pathways Coordinate the Activation of AMPA Receptor-Auxiliary Subunit Complexes. Dawe GB, Musgaard M, Aurousseau MR, Nayeem N, Green T, Biggin PC, Bowie D. Neuron. 2016 Mar 16;89(6):1264-76. doi: 10.1016/j.neuron.2016.01.038. Covered by 7 news outlets: I. Health Canal (http://www.healthcanal.com/brain-nerves/70646-what-makes-the-brain-tick-so-fast.html) II. Health Medicine Network (http://healthmedicinet.com/i/scientists-reveal-how-the-brain-processes-information-with-lightning-speed/) III. News Medical (http://www.news-medical.net/news/20160227/Scientists-reveal-how-the-brain-processes-information-with-lightning-speed.aspx) IV. Wired (http://www.wired.co.uk/article/what-we-learned-about-the-brain-this-week-3) V. Alpha Galileo (http://www.alphagalileo.org/ViewItem.aspx?ItemId=161473&CultureCode=en) VI. Science Daily (https://www.sciencedaily.com/releases/2016/02/160225140254.htm) VII. Eureka Alert (https://www.eurekalert.org/pub_releases/2016-02/mu-wmt022516.php) Recommended by F1000 Prime: http://f1000.com/prime/726180600. 5. Accurate calculation of the absolute free energy of binding for drug molecules. Aldeghi M, Heifetz A, Bodkin MJ, Knapp S, Biggin PC. Chem Sci. 2016 Jan 14;7(1):207-218. Arianna Fornili (Biological and Chemical Sciences, Queen Mary University of London) Publication: Fornili, E. Rostkova, F. Fraternali, M. Pfuhl: Effect of RlC N-Terminal Tails on the Structure and Dynamics of Cardiac Myosin. Biophysical J. 110 (2016) 297A. More in preparation. Dmitry Nerukh (Engineering and Applied Science, Aston University) Important research findings: Distribution of ions inside a viral capsids influences the stability of the capsid Edina Rosta (Computational Chemistry, Kings College London) Highlighted publication: We have a joint experimental paper with the Vertessy group in JACS where we identified a novel arginine finger residue in dUTPase enzymes, and described the mechanism of action for this residue using crystallographic data, biochemical experiments, MD and QM/MM simulations thanks to HECBioSim. http://pubs.acs.org/doi/abs/10.1021/jacs.6b09012 Press release: Our joint paper with Jeremy Baumberg's group in Cambridge was published in Nature. http://www3.imperial.ac.uk/newsandeventspggrp/imperialcollege/newssummary/news_13-6-2016-16-52-4 Greater scientific productivity: As well as speed increases, the optimisation of codes for the ARCHER machine will enable problems to be solved in less time using fewer compute resources. • For the reporting period please provide a brief update on the progress of software development activities associated with the Consortium and the impact this has had on Consortium members and the broader research community. Our lead software development project between Nottingham (Charlie Laughton and Gareth Shannon) and Daresbury (James Gebbie), we have developed a remote job submission tool, 'Longbow' (see below). The tool is designed to reproduce the look and feel of local MD packages, but to stage data and submit jobs to a large HPC resource such as ARCHER in a manner invisible to the user. An open beta version was released unrestricted to the community in March 2015 (http://www.hecbiosim.ac.uk/longbow and via PyPI https://pypi.python.org ) with ~100 downloads. Functionality includes native support for biosimulation packages AMBER, CHARMM, GROMACS, LAMMPS and NAMD, native support for jobs on ARCHER, native support for jobs running on PBS and LSF schedulers, and support for three different job types (single, ensembles and multiple jobs). The software is written both as an application for users and an API for developers. CCP-EM have integrated Longbow into their developmental GUI, and shortly FESetup will ship with native support for job submission using Longbow. We are currently in talks with ClusterVision with respect to them distributing Longbow to their user base as a user friendly way for novices to interact with their systems. Longbow Longbow has been growing in popularity both amongst researchers within the biosimulation field and those in other fields. In this reporting period there have been five new releases of Longbow delivering a plethora of user requested features and bug fixes. Some of the key developments are summarised below: • Implemented a Recovery mode - Should the event happen that Longbow crashes or the system in which Longbow is controlling jobs from powers down. The user can now reconnect to the crashed session and carry on as if nothing happened. • Sub-Queuing - More and more system administrators are setting limits not just on the number of simulations that can run, but also on the number of jobs that can go into the queue. Longbow can now automatically detect this and implement its own queue feeding jobs into the system queue as slots open up. • Dis-connectible/Re-connectible Longbow sessions - A user can now launch Longbow to fire off all jobs and then disconnect, at a later date the user can re-establish the connection and download all results (no need for persistent connections anymore) • Ability to include scripts in the Longbow generated submit files. • Numerous stability, performance and bug fixes Longbow continues to be the submission engine that is used under the hood of the CCP-EM toolkit FLEX-EM. There are several other interesting projects that are making use of Longbow. A project by the energy efficient computing group (Hartree Center) are currently integrating Longbow with CK, a crowd sourced compilation optimisation tool aimed at finding efficient compilation configuration. A project by the Applications performance engineering group (Hartree Center) are currently integrating longbow into Melody, a runtime optimisation tool aimed at optimising runtime configuration of simulations. A project by the computational biology group (Hartree Center) an automated setup, launch and analysis platform for MD on protein-membrane simulations. Metrics Downloads (HECBioSim) 985 Downloads (PyPi) 3456 Increasing the UK's CSE skills base (including graduate and post doctorate training and support): This builds on the skills sets of trained people in HPC, both in terms of capacity and raising the overall skill level available to the sector. • For the reporting period please provide information on the number of PhDs and Post-Docs that have been trained in the use of ARCHER as a result of work relating to the Consortium. • For the reporting period please provide a bullet pointed list of training activities undertaken by the Consortium, providing information on the target audience and level of attendance. 125 PhD and PDRAs have been trained and have used ARCHER through HECBioSim. In the past 6 months alone that has included: PhD students: Marc Dämgen, University of Oxford Laura Domicevica, University of Oxford Matteo Aldeghi, University of Oxford Shaima Hashem, Queen Mary University of London Ruth Dingle, University College London Lucas Siemons, University College London Elvira Tarasova, Aston University Vladimiras Oleinikovas, University College London Havva Yalinca, University College London Daniel Moore, Queen's University Belfast Aaron Maguire, Queen's University Belfast Maxime Tortora, University of Oxford Michail Palaiokostas, UWE Bristol Wei Ding, UWE Bristol Ganesh Shahane, UWE Bristol Pin-Chia Hsu, University of Southampton Damien Jefferies, University of Southampton PDRA: Maria Musgaard, University of Oxford Teresa Paramo, University of Oxford Marieke Schor, University of Edinbrurgh Antonia Mey, University of Edinbrurgh Ioanna Styliari, University of Nottingham Ivan Korotkin, Aston University Silvia Gomez Coca, King's College London Giorgio Saladino, University College London Federico Comitani, University College London Robin Corey, University of Bristol Jordi Juarez-Jimenez, University of Edinburgh Predrag Kukic, University of Cambridge Giulia Tomba, University of Cambridge Deborah Shoemark, University of Bristol Georgios Dalkas, Heriot-Watt University Robin Westacott, Heriot-Watt University Firdaus Samsudin, University of Southampton Agnes Noy, University of Leeds (now EPSRC fellow at York). Please see below details of the training week held in June 2016 at the University of Bristol. 170 delegates attending the training which ran over 5 days. There is course content available on the HECBioSim Workshop Page CCPBioSim Training Week - Day 5: QM/MM enzyme reaction modelling - on Friday 10 June 2016 CCPBioSim Training Week - Day 4: Monte Carlo Methods for Biomodelling - on Thursday 09 June 2016 CCPBioSim Training Week - Day 3: Python for Biomodellers and FESetup - on Wednesday 08 June 2016 CCPBioSim Training Week - Day 2: Running and analysing MD simulations - on Tuesday 07 June 2016 CCPBioSim Training Week - Day 1: Enlighten: Tools for enzyme-ligand modelling - on Monday 06 June 2016 Past Workshops and Conferences: AMOEBA advanced potential energies workshop - on Friday 09 December 2016 Intel Training Workshop (Parallel programming and optimisation for Intel architecture) - on Wednesday 30 November 2016 3rd Workshop on High-Throughput Molecular Dynamics (HTMD) 2016 - on Thursday 10 November 2016 Free Energy Calculation and Molecular Kinetics Workshop - on Tuesday 13 September 2016 Going Large: tools to simplify running and analysing large-scale MD simulations on HPC resources - HECBioSim - on Wednesday 16 December 2015 This 1 day workshop dealt with Longbow, a Python tool created by HECBioSim consortium that allows use of molecular dynamics packages (AMBER, GROMACS, LAMMPS, NAMD) with ease from the comfort of the desktop, and pyPcazip, a flexible Python-based package for the analysis of large molecular dynamics trajectory data sets. Richard Henchman ( Chemistry, Manchester) I lectured on the CCP5 Summer School in 2015 in Manchester and 2016 in Lancaster for the Advanced Topic Biomolecular Simulation drawing on CCPBiosim training materials. Sarah Harris (Physics, Leeds) Agnes Noy Agnes Noy was trained to use ARCHER independently which helped her to obtain an Early-career EPSRC fellowship. The fellowship grant follows the study of supercoiling DNA by modelling (further funding) and introduces a new member to the community (grant is EPSRC (EP/N027639/1)). Increased impact and collaboration with industry: ARCHER does not operate in isolation and the 'impact' of ARCHER's science is converted to economic growth through the interfaces with business and industry. In order to capture the impacts, which may be economic, social, environmental, scientific or political, various metrics may be utilised. • For the reporting period please provide where possible information on Consortium projects that have been performed in collaboration with industry, this should include: o Details of the companies involved. o Information on the part ARCHER and the Consortium played. o A statement on the impact that the work has / is making. o If relevant, details of any in kind or cash contributions that have been associated with this work. • For the reporting period include a list of Consortium publications that have industrial co-authorship. • For the reporting period please provide details of the any other activities involving industrial participation e.g. activities involving any Industrial Advisory panels, attendance / participation in workshops and Consortium based activities. Examples of industrial engagement through HECBioSim include: Syma Khalid (Computational Biophysics, Southampton) Collaboration with Siewert-Jan Marrink (Netherlands) and Wonpil Im(USA) - paper is being written at the moment - also we have developed and tested the following computational tool as part of this project: http://www.charmm-gui.org/?doc=input/membrane We have signed an NDA with a company developing antibiotics (Auspherix) based on work published in Samsudin et al, Structure, 2016. The company provide us with experimental data and crucially the structures of their potential drug molecules, and we work out how they interact with the bacterial membrane. Press releases https://www.sciencedaily.com/releases/2016/11/161121094122.htm ARCHER time we got from HECBioSim was essential for this project. These are large simulations that are very computationally demanding. Julien Michel (Chemistry, Edinburgh) Industrial collaborations that have benefited from HECBioSim UCB was a project partner of EPSRC-funded research in the Michel lab on modelling binding of ligands to flexible proteins. HECBioSim supported the research via allocation of computing time on ARCHER. The work has been published. Impact of Ser17 phosphorylation on the conformational dynamics of the oncoprotein MDM2 Bueren-Calabuig, J. A. ; Michel, J. Biochemistry, 55 (17), 2500-2509, 2016 Elucidation of Ligand-Dependent Modulation of Disorder-Order Transitions in the Oncoprotein MDM2 Bueren-Calabuig, J. A. ; Michel, J. PLoS Comput. Biol. , 11(6): e1004282, 2015 International activities I represented HECBioSim at the MolSSI conceptualisation workshop in Houston (October 2016). Arianna Fornili (School of Biological and Chemical Sciences, Queen Mary University of London) This project is done in collaboration with experimental (NMR and SAXS) partners at King's College London (Dr. Mark Pfuhl and Dr. Elena Rostkova), with the final aim of providing a complete molecular model of how muscle contraction is regulated in the heart. Francesco Gervasio (Chemistry, UCL) A new collaboration with UCB on developing new approaches to sample cryptic binding sites of pharmaceutical interest was started. UCB co-sponsored a BBSRC-CASE PhD studentships. The code development and its application to interesting targets was made possible by the access to Archer. Phil Biggin (Biochemistry, Oxford) The following has industrial partners as co-authors: Accurate calculation of the absolute free energy of binding for drug molecules, Aldeghi M, Heifetz A, Bodkin MJ, Knapp S, Biggin PC. Chem. Sci., 2016,7, 207-218 doi: 10.1039/C5SC02678D. The GLAS scoring module for gromacs has yet to be implemented but that would also constitute industrial activity. Richard Henchman (Chemistry, Manchester) Industrial Collaboration I have one publication with an industrial co-author (Evotec), which Julien Michel is part of as well. Assessment of hydration thermodynamics at protein interfaces with grid cell theory, G. Gerogiokas, M. W. Y. Southey, M. P. Mazanetz, A. Heifetz, M. Bodkin, R. J. Law, R. H. Henchman and J. Michel, J. Phys. Chem. B, 2016, 120, 10442-10452. Strengthening of UK's international position: The impacts of ARCHER's science extend beyond national borders and most science is delivered through partnerships on a national or international level. • For the reporting period please provide a bullet pointed list of projects that have involved international collaboration. For each example please provide a brief summary of the part that ARCHER and the Consortium have played. • For the reporting period please provide a list of consortium publications with international co-authorship. • For the reporting period please detail any other international activities that the Consortium might be involved in (workshops, EU projects etc.). Examples of HECBioSim international engagement include: Michelle Sahai (Department of Life Sciences, Roehampton) Combined in vitro and in silico approaches to the assessment of stimulant properties of novel psychoactive substances - The case of the benzofuran 5-MAPB. Sahai MA, Davidson C, Khelashvili G, Barrese V, Dutta N, Weinstein H, Opacka-Juffry J. Prog Neuropsychopharmacol Biol Psychiatry. (2016) 75:1-9. doi: 10.1016/j.pnpbp.2016.11.004. co-authors affiliations include: Department of Physiology and Biophysics, Weill Cornell Medical College of Cornell University (WCMC), New York, NY, 10065, USA and HRH Prince Alwaleed Bin Talal Bin Abdulaziz Alsaud Institute of Computational Biomedicine, Weill Cornell Medical College of Cornell University, New York, NY, 10065, USA. Author contribution: CD, MAS, HW and JOJ were responsible for the study concept and design. CD and VB collected and interpreted the voltammetry data. JOJ and ND conducted the ligand binding experiments and ND analysed the data. MAS (ARCHER user) and GK performed the molecular modelling studies and interpreted the findings. CD, MAS, GK and JOJ drafted the manuscript. All authors critically reviewed the content and approved the final version for publication. Dmitry Nerukh (Engineering and Applied Science, Aston University) Collaborations: • Prof. Makoto Taiji group (K-computer, MDGRAPE), Laboratory for Computational Molecular Design, Computational Biology Research Core, RIKEN Quantitative Biology Center (QBiC), Kobe, Japan • Prof. Reza Khayat group, City College of New York, United States • Prof. Artyom Yurov group, Immanuel Kant Baltic Federal University, Russian Federation • Prof. Nikolay Mchedlov-Petrosyan group, V.N. Karazin Kharkiv National University, Ukraine • Prof. Natalya Vaysfeld group, Odessa I.I.Mechnikov National University, Ukraine International activities: • International Workshop - Engineering bacteriophages for treating antimicrobial resistance using computational models. Dec. 2016, Aston University, UK Dmitry Nerukh group at Aston University collaborates with the group of Prof. Makoto Taiji (the deputy director of RIKEN Quantitative Biology Institute). Prof. Taiji group is the author and implementer of the fastest machine in the world for molecular dynamics simulations, MDGRAPE, as well as part of the team designing the K-computer and working on it (several times the fastest computer in the world). We are currently preparing a joint manuscript for publication where the results obtained on ARCHER will be used alongside with the results obtained on MDGRAPE-4. Julien Michel (Chemistry, Edinburgh) Free Energy Reproducibility Project Free energy reproducibility project between the Michel, Mobley, Roux groups and STFC. This makes use of FESetup which is CCPBioSim, but calculations have been executed on various HPC platforms, Hannes can clarify whether any of the systems used is within the remit of HECBioSim. Sarah Harris (Physics, Leeds), Charlie Laughton (Pharmacy, Nottingham) and Agnes Noy (Physics, York) In an international collaboration with the Institute for Research in Biomedicine in Barcelona, Noy, Harris and Laughton used ARCHER time obtained through the HecBioSim consortium to perform multiple 100ns molecular dynamics (MD) simulations of DNA minicircles containing ~100 base pairs as part of the validation suite for the new BSC1 nucleic acid forcefield, see: Ivani I., Dans P. D., Noy A., Perez A., Faustino I., Hospital A., Walther J., Andrio P., Goni R., Portella G., Battistini F.,Gelpi J. L. Gonzalez C., Vendruscolo M., Laughton C. A., Harris S. A. Case D. A. & Orozco M. "Parmbsc1: a refined force field for DNA simulations" Nat. Methods 13, 55, 2015, doi:10.1038/nmeth.3658 Jon Essex (Chemistry, Southampton) Development and testing of hybrid coarse-grain/atomistic model of membrane systems. ARCHER and the consortium were instrumental in providing the computing time necessary to complete this work. The work was performed by a research fellow in my group, who has subsequently taken up an academic post in Sweden, where he is continuing to develop this methodology. Publication: All-atom/coarse-grained hybrid predictions of distributioncoefficients in SAMPL5 Samuel Genheden1, Jonathan W. Essex, J Comput Aided Mol Des (2016) 30:969-976 DOI 10.1007/s10822-016-9926-z IP and other industrial engagement, or translation, which has benefited from HECBioSim international collaborations Collaboration Dr Samuel Genheden, University of Gothenburg - see above International activities (e.g. workshops) that have benefited from HECBioSim The work performed with Sam resulted in an successful eCSE application for development work on the LAMMPS software. This is an international simulation package run out of Sandia labs in the US. Through this eCSE application we were able to update the rotational integrator and improve the load balancing for our hybrid simulation methodology. These developments have been incorporated in the latest release versions of the code. An ARCHER training webinar resulted from this work. Training activities see above - ARCHER training webinar on our developments within LAMMPS Software development see above - code modifications in LAMMPS which either have been, or will be, in the release version of the code Advantages you see of the consortium (e.g. could be ability to pursue project quickly; develop collaborations) resources to run the sort of large-scale calculations we need Michelle Sahai (Department of life Sciences, Roehampton) The publication I added on ResearchFish has international co-authors. Combined in vitro and in silico approaches to the assessment of stimulant properties of novel psychoactive substances - The case of the benzofuran 5-MAPB. Sahai MA, Davidson C, Khelashvili G, Barrese V, Dutta N, Weinstein H, Opacka-Juffry J. Prog Neuropsychopharmacol Biol Psychiatry. (2016) 75:1-9. doi: 10.1016/j.pnpbp.2016.11.004. co-authors affiliations include: Department of Physiology and Biophysics, Weill Cornell Medical College of Cornell University (WCMC), New York, NY, 10065, USA and HRH Prince Alwaleed Bin Talal Bin Abdulaziz Alsaud Institute of Computational Biomedicine, Weill Cornell Medical College of Cornell University, New York, NY, 10065, USA. Author contribution: CD, MAS, HW and JOJ were responsible for the study concept and design. CD and VB collected and interpreted the voltammetry data. JOJ and ND conducted the ligand binding experiments and ND analysed the data. MAS (ARCHER user) and GK performed the molecular modelling studies and interpreted the findings. CD, MAS, GK and JOJ drafted the manuscript. All authors critically reviewed the content and approved the final version for publication. Edina Rosta (Computational Chemistry, Kings College London) International collaborations I have several international collaborators with joint computational/experimental projects. For their success the HECBioSim computer time was essential: • Prof. Beata Vertessy, Budapest Technical University, Hungary JACS 2016 138 (45), 15035-15045 • Profs. Walter Kolch, Vio Buchete and Boris Kholodenko, UCD, Ireland Angewandte 55 (3), 983-986, 2016 (this may have been reported previously) PLOS Computational Biology 12 (10), e1005051, 2016 J Phys Chem Letters 7 (14), 2676-2682, 2016 • Jose Maria Lluch, Angels Gonzalez, Autonomous University of Barcelona, Spain JCTC 12 (4), 2079-2090, 2016 Other international activities Free energy and molecular kinetics workshop with Frank Noe's group (Free University of Berlin) and Vio Buchete (UCD, Dublin). Maybe this should be for training activities? Following the workshop, Erasmus students apply to visit my group. Phil Biggin (Biochemistry, Oxford) 2 and 4 were with colleagues from McGill (Brown PM, Aurousseau MR and Bowie) and 3 was with colleagues from Denmark (Braun and Pless) 2. Kainate receptor pore-forming and auxiliary subunits regulate channel block by a novel mechanism. Brown PM, Aurousseau MR, Musgaard M, Biggin PC, Bowie D. J Physiol. 2016 Apr 1;594(7):1821-40. doi: 10.1113/JP271690. 3. Role of an Absolutely Conserved Tryptophan Pair in the Extracellular Domain of Cys-Loop Receptors. Braun N, Lynagh T, Yu R, Biggin PC, Pless SA. ACS Chem Neurosci. 2016 Mar 16;7(3):339-48. doi: 10.1021/acschemneuro.5b00298. 4. Distinct Structural Pathways Coordinate the Activation of AMPA Receptor-Auxiliary Subunit Complexes. Dawe GB, Musgaard M, Aurousseau MR, Nayeem N, Green T, Biggin PC, Bowie D. Neuron. 2016 Mar 16;89(6):1264-76. doi: 10.1016/j.neuron.2016.01.038. Richard Henchman I have an international collaboration with Professor Franke Grater at the University of Heidelberg on entropy theory for biomolecular systems. Syma Khalid (Chemistry, Southampton) Collaboration with Singapore National Center for Biotechnology. Contributions to CECAM workshops Other Highlights for the Current Reporting Period: Please provide details of any other significant highlights from the reporting period that are not captured elsewhere in the report. Professor Adrian Mullholland organised and chaired the Computational Chemistry, Gordon Research Conference, Girona, Spain 24th - 29th July 2016. The theme of the 2016 Computational Chemistry GRC was "Theory and Simulation Across Scales in Molecular Science". It focused on method development and state-of-the-art applications across computational molecular science, and encouraged cross-fertilization between areas. The conference was oversubscribed, with attendees from industry and academia. The meeting received a High-Performance Rating, and was commended for the assessment that 95% of conferees rated this meeting "above average" on all evaluation areas (science, discussion, management, atmosphere and suitability). HEC Consortia Model: Over the coming months EPSRC will be looking at the future of the HEC Consortia model and potential future funding. We would like to use this opportunity to ask the Consortia Chairs for input: • What are the key benefits that your community have experienced through the existence of the HEC Consortia? • What elements of the financial support provided by the HEC Consortium's grant have worked well and what could be improved in the future? We aim to expand the breadth of the work of the Consortium focusing on cutting-edge applications, and building collaborations with experiments and industry, to achieve maximum impact from ARCHER use. We discussed Grand Challenges at our most recent Management Group meeting. In December 2015, and have identified several to follow up as strategic priorities. We aim to tackle and support large-scale grand challenge applications in biomolecular science, in areas such as antimicrobial resistance, membrane dynamics, drug design and synthetic biology. One specific theme with potentially high impact in drug discovery is large -scale comparative investigation of allosteric regulation in different superfamilies of proteins (e.g. PAS domain containing proteins, tyrosine kinases, etc.). The major impact will be in the identification of novel selective therapeutic molecules with limited adverse side effects. As a Consortium, we intend to develop and apply novel computational approaches for the rational design of allosteric regulators of hitherto 'undruggable' targets. Many of the targets emerging from large-scale genetic screening are deemed undruggable due to the difficulty of designing drug-like modulators that bind to their catalytic sites. It is increasingly clear that these targets might be effectively targeted by designing drugs that bind to protein-protein interfaces and allosteric sites. To do that, however, new rational design strategies, based on an in-depth knowledge of protein dynamics and advanced modelling and new simulation techniques are required. Other challenging frontier areas include dynamics of motor proteins; prediction of the effects of pathogenic mutations on protein function. The accurate prediction of free energy of binding is still in its infancy and needs much further investigation. Finally, kinetics of biomolecular reactions will become the next big topic. It is clear that high end computing resource can provide the necessary power and capability to provide inroads into this vital area. This is important because it is becoming increasingly apparent that kinetics of drug binding is a key factor that the pharmaceutical sector should really be looking at in terms of a major dictator of potency. HECBioSim is now well established, supporting work of many groups across the UK in the growing field of biomolecular simulation. We benefit from an Advisory Group containing members from industry, as we as international biomolecular simulation experts and experimental scientists. The Advisory Board was expanded and refreshed for the renewal and consists of: Dr. Nicolas Foloppe (Vernalis plc, Chair); Dr. Colin Edge (GlaxoSmithKline); Dr. Mike King (UCB Pharmaceuticals); Dr. Mike Mazanetz (Evotec AG); Dr. Garrett Morris (Crysalin Ltd.); Dr. Gary Tresadern (Johnson and Johnson Pharmaceuticals); Dr. Richard Ward (AstraZeneca); Prof. Modesto Orozco (IRB, Barcelona); Prof. Tony Watts, (Oxford, NMR); Dr. Pete Bond (A*STAR Bioinformatics Institute Singapore). We aim to foster industrial collaborations and collaborations with experimentalists (e.g. joint workshops with CCPN, Institute of Physics (Sarah Harris, Leeds Physics); see e.g. case studies. A particular theme for strategic development will be multiscale modelling, building on collaborations between several groups in the Consortium and the other CCPs. This theme reflects the inclusive and forward looking philosophy of HecBioSim, which is a community open to new ideas, keen to develop new methods, and ultimately to use HEC to drive exciting new science. The Consortium model allows new collaborations to develop. The recent reduction in time allocated to HECBioSim has meant that we can support fewer projects. There is significantly more demand for computer time than we can accommodate through our current allocation. We intend to work with Tier 2 Centres to explore possibilities for applications, and e.g. to test emerging architectures for biomolecular simulation. Edina Rosta (Computational Chemistry, Kings College London) comments: "HECBioSim enables my group to perform biomolecular simulations at the high standards required for JACS, Angewandte, etc. Without this consortium I would not be able to perform the necessary calculations leading to publishable work in top journals as our local university resources are limited and the hpc systems are not well maintained". Regarding financial support: the administrative support provided via the consortium grant is essential to the functioning of HECBioSim. This role is currently undertaken by Simone Breckell who not only administers, allocates time to projects and arranges panel meetings but has also done a fantastic job collating the extensive information for this EPSRC report. |
| URL | http://www.hecbiosim.ac.uk |
| Title | COVID is airborne |
| Description | Multscale model of SARS-CoV-2 in respiratory aerosol |
| Type Of Material | Improvements to research infrastructure |
| Year Produced | 2022 |
| Provided To Others? | Yes |
| Impact | HPC2021 Gordon Bell Special Prize Finalist Skip to main content U.S. flagAn official website of the United States government Here's how you know NIH NLM Logo Log in Access keysNCBI HomepageMyNCBI HomepageMain ContentMain Navigation Search PMC Full-Text Archive Search in PMC Run this search in PubMed Advanced Search User Guide Journal List SAGE Choice PMC9527558 Logo of sageopen The International Journal of High Performance Computing Applications Int J High Perform Comput Appl. 2023 Jan; 37(1): 28-44. Published online 2022 Oct 2. doi: 10.1177/10943420221128233 PMCID: PMC9527558 PMID: 36647365 #COVIDisAirborne: AI-enabled multiscale computational microscopy of delta SARS-CoV-2 in a respiratory aerosol Monitoring Editor: Mark Parsons Abigail Dommer,1,†Lorenzo Casalino,1,†Fiona Kearns,1,†Mia Rosenfeld,1 Nicholas Wauer,1 Surl-Hee Ahn,1 John Russo,2 Sofia Oliveira,3 Clare Morris,1 Anthony Bogetti,4 Anda Trifan,5,6 Alexander Brace,5,7 Terra Sztain,1,8 Austin Clyde,5,7 Heng Ma,5 Chakra Chennubhotla,4 Hyungro Lee,9 Matteo Turilli,9 Syma Khalid,10 Teresa Tamayo-Mendoza,11 Matthew Welborn,11 Anders Christensen,11 Daniel GA Smith,11 Zhuoran Qiao,12 Sai K Sirumalla,11 Michael O'Connor,11 Frederick Manby,11 Anima Anandkumar,12,13 David Hardy,6 James Phillips,6 Abraham Stern,13 Josh Romero,13 David Clark,13 Mitchell Dorrell,14 Tom Maiden,14 Lei Huang,15 John McCalpin,15 Christopher Woods,3 Alan Gray,13 Matt Williams,3 Bryan Barker,16 Harinda Rajapaksha,16 Richard Pitts,16 Tom Gibbs,13 John Stone,6,13 Daniel M. Zuckerman,2 Adrian J. Mulholland,3 Thomas Miller, III,11,12 Shantenu Jha,9 Arvind Ramanathan,5 Lillian Chong,4 and Rommie E Amaro1 Author information Copyright and License information Disclaimer Previous version available: This article is based on a previously available preprint: "#COVIDisAirborne: AI-Enabled Multiscale Computational Microscopy of Delta SARS-CoV-2 in a Respiratory Aerosol". Go to: Abstract We seek to completely revise current models of airborne transmission of respiratory viruses by providing never-before-seen atomic-level views of the SARS-CoV-2 virus within a respiratory aerosol. Our work dramatically extends the capabilities of multiscale computational microscopy to address the significant gaps that exist in current experimental methods, which are limited in their ability to interrogate aerosols at the atomic/molecular level and thus obscure our understanding of airborne transmission. We demonstrate how our integrated data-driven platform provides a new way of exploring the composition, structure, and dynamics of aerosols and aerosolized viruses, while driving simulation method development along several important axes. We present a series of initial scientific discoveries for the SARS-CoV-2 Delta variant, noting that the full scientific impact of this work has yet to be realized. Keywords: molecular dynamics, deep learning, multiscale simulation, weighted ensemble, computational virology, SARS-CoV-2, aerosols, COVID-19, HPC, AI, GPU, Delta Go to: Justification We develop a novel HPC-enabled multiscale research framework to study aerosolized viruses and the full complexity of species that comprise them. We present technological and methodological advances that bridge time and length scales from electronic structure through whole aerosol particle morphology and dynamics. Performance attributes Performance attribute Our submission Category of achievement Scalability, Time-to-solution Type of method used Explicit, Deep Learning Results reported on the basis of Whole application including I/O Precision reported Mixed Precision System scale Measured on full system Measurement mechanism Hardware performance counters Application timers Performance Modeling Open in a separate window Overview of the problem Respiratory pathogens, such as SARS-CoV-2 and influenza, are the cause of significant morbidity and mortality worldwide. These respiratory pathogens are spread by virus-laden aerosols and droplets that are produced in an infected person, exhaled, and transported through the environment (Wang et al., 2021) (Figure 1). Medical dogma has long focused on droplets as the main transmission route for respiratory viruses, where either a person has contact with an infected surface (fomites) or direct droplet transmission by close contact with an infected individual. However, as we continue to observe with SARS-CoV-2, airborne transmission also plays a significant role in spreading disease. We know this from various super spreader events, for example, during a choir rehearsal (Miller et al., 2021). Intervention and mitigation decisions, such as the relative importance of surface cleaning or whether and when to wear a mask, have unfortunately hinged on a weak understanding of aerosol transmission, to the detriment of public health. An external file that holds a picture, illustration, etc. Object name is 10.1177_10943420221128233-fig1.jpg Figure 1. Overall schematic depicting the construction and multiscale simulations of Delta SARS-CoV-2 in a respiratory aerosol. (N.B.: The size of di-valent cations has been increased for visibility.) A central challenge to understanding airborne transmission has been the inability of experimental science to reliably probe the structure and dynamics of viruses once they are inside respiratory aerosol particles. Single particle experimental methods have poor resolution for smaller particles (<1 micron) and are prone to sample destruction during collection. Airborne viruses are present in low concentrations in the air and are similarly prone to viral inactivation during sampling. In addition, studies of the initial infection event, for example, in the deep lung, are limited in their ability to provide a detailed understanding of the myriad of molecular interactions and dynamics taking place in situ. Altogether, these knowledge gaps hamper our collective ability to understand mechanisms of infection and develop novel effective antivirals, as well as prevent us from developing concrete, science-driven mitigation measures (e.g., masking and ventilation protocols). Here, we aim to reconceptualize current models of airborne transmission of respiratory viruses by providing never-before-seen views of viruses within aerosols. Our approach relies on the use of all-atom molecular dynamics (MD) simulations as a multiscale "computational microscope." MD simulations can synthesize multiple types of biological data (e.g., multiresolution structural datasets, glycomics, lipidomics, etc.) into cohesive, biologically "accurate" structural models. Once created, we then approximate the model down to its many atoms, creating trajectories of its time dependent dynamics under cell-like (or in this case, aerosol-like) conditions. Critically, MD simulations are more than just "pretty movies." MD equations are solved in a theoretically rigorous manner, allowing us to compute experimentally testable macroscopic observables from time-averaged microscopic properties. What this means is that we can directly connect MD simulations with experiments, each validating and providing testable hypotheses to the other, which is the real power of the approach. An ongoing challenge to the successful application of such methods, however, is the need for technological and methodological advances that make it possible to access length scales relevant to the study of large, biologically complex systems (spanning nanometers to ~one micron in size) and, correspondingly, longer timescales (microseconds to seconds). Such challenges and opportunities manifest in the study of aerosolized viruses. Aerosols are generally defined as being less than 5 microns in diameter, able to float in the air for hours, travel significant distances (i.e., can fill a room, like cigarette smoke), and be inhaled. Fine aerosols < 1 micron in size can stay in the air for over 12 h and are enriched with viral particles (Fennelly 2020; Coleman et al., 2021). Our work focuses on these finer aerosols that travel deeper into the respiratory tract. Several studies provide the molecular recipes necessary to reconstitute respiratory aerosols according to their actual biologically relevant composition (Vejerano and Marr 2018; Walker et al., 2021). These aerosols can contain lipids, cholesterol, albumin (protein), various mono- and di-valent salts, mucins, other surfactants, and water (Figure 1). Simulations of aerosolized viruses embody a novel framework for the study of aerosols: they will allow us and others to tune different species, relative humidity, ion concentrations, etc. to match experiments that can directly and indirectly connect to and inform our simulations, as well as test hypotheses. Some of the species under study here, for example, mucins, have not yet been structurally characterized or explored with simulations and thus the models we generate are expected to have impact beyond their roles in aerosols. In addition to varying aerosol composition and size, the viruses themselves can be modified to reflect new variants of concern, where such mutations may affect interactions with particular species in the aerosol that might affect its structural dynamics and/or viability. The virion developed here is the Delta variant (B.1.617.2 lineage) of SARS-CoV-2 (Figure 2), which presents a careful integration of multiple biological datasets: (1) a complete viral envelope with realistic membrane composition, (2) fully glycosylated full-length spike proteins integrating 3D structural coordinates from multiple cryoelectron microscopy (cryoEM) studies (McCallum et al., 2021; Wrapp et al., 2020; Walls et al., 2020; Bangaru et al., 2020) (3) all biologically known features (post-translational modifications, palmitoylation, etc.), (4) any other known membrane proteins (e.g. the envelope (E) and membrane (M) proteins), and (5) virion size and patterning taken directly from cryoelectron tomography (cryoET). Each of the individual components of the virus are built up before being integrated into the composite virion, and thus represent useful molecular-scale scientific contributions in their own right (Casalino et al., 2020; Sztain et al., 2021). An external file that holds a picture, illustration, etc. Object name is 10.1177_10943420221128233-fig2.jpg Figure 2. Individual protein components of the SARS-CoV-2 Delta virion. The spike is shown with the surface in cyan and with Delta's mutated residues and deletion sites highlighted in pink and yellow, respectively. Glycans attached to the spike are shown in blue. The E protein is shown in yellow and the M-protein is shown in silver and white. Visualized with VMD. Altogether in this work, we dramatically extend the capabilities of data-driven, multiscale computational microscopy to provide a new way of exploring the composition, structure, and dynamics of respiratory aerosols. While a seemingly limitless number of putative hypotheses could result from these investigations, the first set of questions we expect to answer are: How does the virus exist within a droplet of the same order of magnitude in size, without being affected by the air-water interface, which is known to destroy molecular structure (D'Imprima et al. 2019)? How does the biochemical composition of the droplet, including pH, affect the structural dynamics of the virus? Are there species within the aerosols that "buffer" the viral structure from damage, and are there particular conditions under which the impact of those species changes? Our simulations can also provide specific parameters that can be included in physical models of aerosols, which still assume a simple water or water-salt composition even though it is well known that such models, for example, using kappa-Kohler theory, break down significantly as the molecular species diversify (Petters and Kreidenweis 2007). Go to: Current state of the art Current experimental methods are unable to directly interrogate the atomic-level structure and dynamics of viruses and other molecules within aerosols. Here we showcase computational microscopy as a powerful tool capable to overcome these significant experimental limitations. We present the major elements of our multiscale computational microscope and how they come together in an integrated manner to enable the study of aerosols across multiple scales of resolution. We demonstrate the impact such methods can bring to bear on scientific challenges that until now have been intractable, and present a series of new scientific discoveries for SARS-CoV-2. Parallel molecular dynamics All-atom molecular dynamics simulation has emerged as an increasingly powerful tool for understanding the molecular mechanisms underlying biophysical behaviors in complex systems. Leading simulation engines, NAMD (Phillips et al., 2020), AMBER (Case et al. [n. d.]), and GROMACS (Páll et al., 2020), are broadly useful, with each providing unique strengths in terms of specific methods or capabilities as required to address a particular biological question, and in terms of their support for particular HPC hardware platforms. Within the multiscale computational microscopy platform developed here, we show how each of these different codes contributes different elements to the overall framework, oftentimes utilizing different computing modalities/architectures, while simultaneously extending on state-of-the-art for each. Structure building, simulation preparation, visualization, and post hoc trajectory analysis are performed using VMD on both local workstations and remote HPC resources, enabling modeling of the molecular systems studied herein (Humphrey et al., 1996; Stone et al., 2013a,b, 2016b; Sener et al., 2021). We show how further development of each of these codes, considered together within the larger-scale collective framework, enables the study of SARS-CoV-2 in a wholly novel manner, with extension to numerous other complex systems and diseases. AI-enhanced WE simulations Because the virulence of the Delta variant of SARS-CoV-2 may be partly attributable to spike protein (S) opening, it is of pressing interest to characterize the mechanism and kinetics of the process. Although S-opening in principle can be studied via conventional MD simulations, in practice the system complexity and timescales make this wholly intractable. Splitting strategies that periodically replicate promising MD trajectories, among them the weighted ensemble (WE) method (Huber and Kim 1996; Zuckerman and Chong 2017), have enabled simulations of the spike opening of WT SARS-CoV-2 (Sztain et al., 2021; Zimmerman et al., 2021). WE simulations can be orders of magnitude more efficient than conventional MD in generating pathways and rate constants for rare events (e.g. protein folding (Adhikari et al., 2019) and binding (Saglam and Chong 2019)). The WESTPA software for running WE (Zwier et al., 2015) is well-suited for high-performance computing with nearly perfect CPU/GPU scaling. The software is interoperable with any dynamics engine, including the GPU-accelerated AMBER dynamics engine (Salomon-Ferrer et al., 2013) that is used here. As shown below, major upgrades to WESTPA (v. 2.0) have enabled a dramatic demonstration of spike opening in the Delta variant (Figures 5 and and6)6) and exponentially improved analysis of spike-opening kinetics (Russo et al., 2022). An external file that holds a picture, illustration, etc. Object name is 10.1177_10943420221128233-fig5.jpg Figure 5. Delta variant spike opening from WE simulations, and AI/haMSM analysis. A) The integrated workflow. B) Snapshots of the "down," "up," and "open" states for Delta S-opening from a representative pathway generated by WE simulation, which represents ~ 105 speedup compared to conventional MD. C) Rate constant estimation with haMSM analysis of WE data (purple lines) significantly improves direct WE computation (red), by comparison to experimental measurement (black dashed). Varying haMSM estimates result from different featurizations which will be individually cross-validated. D) The first three dimensions of the ANCA-AE embeddings depict a clear separation between the closed (darker purple) and open (yellow) conformations of the Delta spike. A sub-sampled landscape is shown here where each sphere represents a conformation from the WE simulations and colored with the root-mean squared deviations (Å) with respect to the closed state. Visualized with VMD. An external file that holds a picture, illustration, etc. Object name is 10.1177_10943420221128233-fig6.jpg Figure 6. WE simulations reveal a dramatic opening of the Delta S (cyan), compared to WT S (white). While further investigation is needed, this super open state seen in the Delta S may indicate increased capacity for binding to human host-cell receptors. The integration of AI techniques with WE can further enhance the efficiency of sampling rare events (Noe 2020; Brace et al., 2021b; Casalino et al., 2021). One frontier area couples unsupervised linear and non-linear dimensionality reduction methods to identify collective variables/progress coordinates in high-dimensional molecular systems (Bhowmik et al., 2018; Clyde et al., 2021). Such methods may be well suited for analyzing the aerosolized virus. Integrating these approaches with WE simulations is advantageous in sampling the closed ? open transitions in the Delta S landscape (Figure 5) as these unsupervised AI approaches automatically stratify progress coordinates (Figure 5(D)). Dynamical non-equilibrium MD Aerosols rapidly acidify during flight via reactive uptake of atmospheric gases, which is likely to impact the opening/closing of the S protein (Vejerano and Marr 2018; Warwicker 2021). Here, we describe the extension of dynamical non-equilibrium MD (D-NEMD) (Ciccotti and Ferrario 2016) to investigate pH effects on the Delta S. D-NEMD simulations (Ciccotti and Ferrario 2016) are emerging as a useful technique for identifying allosteric effects and communication pathways in proteins (Galdadas et al., 2021; Oliveira et al., 2019), including recently identifying effects of linoleic acid in the WT spike (Oliveira et al., 2021b). This approach complements equilibrium MD simulations, which provide a distribution of configurations as starting points for an ensemble of short non-equilibrium trajectories under the effect of the external perturbation. The response of the protein to the perturbation introduced can then be determined using the Kubo-Onsager relation (Oliveira et al., 2021a; Ciccotti and Ferrario 2016) by directly tracking the change in atomic positions between the equilibrium and non-equilibrium simulations at equivalent points in time (Oliveira et al., 2021a). OrbNet Ca2+ ions are known to play a key role in mucin aggregation in epithelial tissues (Hughes et al., 2019). Our RAV simulations would be an ideal case-study to probe such complex interactions between Ca2+, mucins, and the SARS-CoV-2 virion in aerosols. However, Ca2+ binding energies can be difficult to capture accurately due to electronic dispersion and polarization, terms which are not typically modeled in classical mechanical force fields. Quantum mechanical (QM) methods are uniquely suited to capture these subtle interactions. Thus, we set out to estimate the correlation in Ca2+ binding energies between CHARMM36m and quantum mechanical estimates enabled via AI with OrbNet. Calculation of energies with sufficient accuracy in biological systems can, in many cases, be adequately described with density functional theory (DFT). However, its high cost limits the applicability of DFT in comparison to fixed charge force fields. To capture quantum quality energetics at a fraction of the computational expense, we employ a novel approach (OrbNet) based on the featurization of molecules in terms of symmetry-adapted atomic orbitals and the use of graph neural network methods for deep learning quantum-mechanical properties (Qiao et al., 2020). Our method outperforms existing methods in terms of its training efficiency and transferable accuracy across diverse molecular systems, opening a new pathway for replacing DFT in large-scale scientific applications such as those explored here. (Christensen et al., 2021). Innovations realized Construction and simulation of SARS-CoV-2 in a respiratory aerosol Our approach to simulating the entire aerosol follows a composite framework wherein each of the individual molecular pieces is refined and simulated on its own before it is incorporated into the composite model. Simulations of each of the components are useful in their own right, and often serve as the basis for biochemical and biophysical validation and experiments (Casalino et al., 2020). Throughout, we refer to the original circulating SARS-CoV-2 strain as "WT," whereas all SARS-CoV-2 proteins constructed in this work represent the Delta variant (Figure 2). All simulated membranes reflect mammalian ER-Golgi intermediate compartment (ERGIC) mimetic lipid compositions. VMD (Humphrey et al., 1996; Stone et al., 2016a), psfgen (Phillips et al., 2005), and CHARMM-GUI (Park et al., 2019) were used for construction and parameterization. Topologies and parameters for simulations were taken from CHARMM36m all-atom additive force fields (Guvench et al., 2009; Huang and Mackerell 2013; Huang et al., 2017; Klauda et al., 2010; Beglov and Roux 1994; Han et al., 2018; Venable et al., 2013). NAMD was used to perform MD simulations (Phillips et al., 2020), adopting similar settings and protocols as in (Casalino et al., 2020). All systems underwent solvation, charge neutralization, minimization, heating, and equilibration prior to production runs. Refer to Table 1 for Abbreviations, PBC dimensions, total number of atoms, and total equilibration times for each system of interest. Table 1. Summary of all systems constructed in this work. See Figure 3 for illustration of aerosol construction. asystems bAbb c(Å × Å × Å) dNa e (ns) fM dimers M 125 × 125 × 124 164,741 700 fE pentamers E 123 × 125 × 102 136,775 41 Spikes f (Open) S 206 × 200 × 410 1,692,444 330 f (Closed) S 204 × 202 × 400 1,658,224 330 g (Closed, head) SH 172 × 184 × 206 615,593 73µs Mucins fshort mucin 1 m1 123 × 104 × 72 87,076 25 fshort mucin 2 m2 120 × 101 × 72 82,155 25 flong mucin 1 m3 810 × 104 × 115 931,778 23 flong mucin 2 m4 904 × 106 × 109 997,029 15 flong mucin 3 m5 860 × 111 × 113 1,040,215 18 fS+m1/m2+ALB SMA 227 × 229 × 433 2,156,689 840 fVirion V 1460 × 1460 × 1460 305,326,834 41 fResp.Aero.+Vir RAV 2834 × 2820 × 2828 1,016,813,441 2.42 Total FLOPS 2.4 ZFLOPS Open in a separate window aM, E, S, SH, and V models represent SARS-CoV-2 Delta strain. bAbbreviations used throughout document. cPeriodic boundary dimensions. dTotal number of atoms. eTotal aggregate simulation time, including heating and equilibration runs. fSimulated with NAMD. gSimulated with NAMD, AMBER, and GROMACS. Simulating the SARS-CoV-2 structural proteins Fully glycosylated Delta spike (S) structures in open and closed conformations were built based on WT constructs from Casalino et al. (Casalino et al., 2020) with the following mutations: T19R, T95I, G142D, E156G, ?157-158, L452R, T478K, D614G, P681R, and D950N (McCallum et al., 2021; Kannan et al., 2021). Higher resolved regions were grafted from PDB 7JJI (Bangaru et al., 2020). Additionally, coordinates of residues 128-167-accounting for a drastic conformational change seen in the Delta variant S-graciously made available to us by the Veesler Lab, were similarly grafted onto our constructs (McCallum et al., 2021). Finally, the S proteins were glycosylated following work by Casalino et al. (Casalino et al., 2020). By incorporating the Veesler Lab's bleeding-edge structure (McCallum et al., 2021) and highly resolved regions from 7JJI (Bangaru et al., 2020), our models represent the most complete and accurate structures of the Delta S to date. The S proteins were inserted into membrane patches and equilibrated for 3 × 110 ns. For non-equilibrium and weighted ensemble simulations, a closed S head (SH, residues 13-1140) was constructed by removing the stalk from the full-length closed S structure, then resolvated, neutralized, minimized, and subsequently passed to WE and D-NEMD teams. The M-protein was built from a structure graciously provided by the Feig Lab (paper in prep). The model was inserted into a membrane patch and equilibrated for 700 ns. RMSD-based clustering was used to select a stable starting M-protein conformation. From the equilibrated and clustered M structure, VMD's Mutator plugin (Humphrey et al., 1996) was used to incorporate the I82T mutation onto each M monomer to arrive at the Delta variant M. To construct the most complete E protein model to-date, the structure was patched together by resolving incomplete PDBs 5X29 (Surya et al., 2018), 7K3G (Mandala et al., 2020) and 7M4R (Chai et al., 2021). To do so, the transmembrane domain (residues 8-38) from 7K3G were aligned to the N-terminal domain (residues 1-7) and residues 39 to 68 of 5X29 and residues 69 to 75 of 7M4R by their Ca atoms. E was then inserted into a membrane patch and equilibrated for 40 ns. Constructing the SARS-CoV-2 Delta virion The SARS-CoV-2 Delta virion (V) model was constructed following Casalino et al. (Casalino et al., 2021) using CHARMM-GUI (Lee et al., 2016), LipidWrapper (Durrant and Amaro 2014), and Blender (Blender Online Community 2020), using a 350 Å lipid bilayer with an equilibrium area per lipid of 63 Å2 and a 100 nm diameter Blender icospherical surface mesh (Turonova et al., 2020). The resulting lipid membrane was solvated in a 1100 Å3 waterbox and subjected to four rounds of equilibration and patching (Casalino et al., 2021). 360 m dimers and 4 E pentamers were then tiled onto the surface, followed by random placement of 29 full-length S proteins (9 open, 20 closed) according to experimentally observed S protein density (Ke et al., 2020). M and E proteins were oriented with intravirion C-termini. After solvation in a 1460 Å waterbox, the complete V model tallied >305 million atoms (Table 1). V was equilibrated for 41 ns prior to placement in the respiratory aerosol (RA) model. The equilibrated membrane was 90 nm in diameter and remains in close structural agreement with the experimental studies (Ke et al., 2020). Building and simulating the respiratory aerosol Respiratory aerosols contain a complex mixture of chemical and biological species. We constructed a respiratory aerosol (RA) fluid based on a composition from artificial saliva and surrogate deep lung fluid recipes (Walker et al., 2021). This recipe includes 0.7 mm DPPG, 6.5 mm DPPC, 0.3 mm cholesterol, 1.4 mm Ca2+, 0.8 mm Mg2+, and 142 mm Na+ (Vejerano and Marr 2018; Walker et al., 2021), human serum albumin (ALB) protein, and a composition of mucins (Figure 3). Mucins are long polymer-like structures that are decorated by dense, heterogeneous, and complex regions of O-glycans. This work represents the first of its kind as, due to their complexity, the O-glycosylated regions of mucins have never before been constructed for molecular simulations. Two short (m1, m2, ~5 nm) and three long (m3, m4, m5 ~55 nm) mucin models were constructed following known experimental compositions of protein and glycosylation sequences (Symmes et al., 2018; Hughes et al., 2019; Markovetz et al., 2019; Thomsson et al., 2005; Mariethoz et al., 2018) with ROSETTA (Raveh et al., 2010) and CHARMM-GUI Glycan Modeller (Jo et al., 2011). Mucin models (short and long) were solvated, neutralized by charge matching with Ca2+ ions, minimized, and equilibrated for 15-25 ns each (Table 1). Human serum albumin (ALB), which is also found in respiratory aerosols, was constructed from PDB 1AO6 (Sugio et al., 1999). ALB was solvated, neutralized, minimized, and equilibrated for 7ns. Equilibrated structures of ALB and the three long mucins were used in construction of the RAV with m3+m4+m5 added at 6 g/mol and ALB at 4.4 g/mol. An external file that holds a picture, illustration, etc. Object name is 10.1177_10943420221128233-fig3.jpg Figure 3. Image of RAV with relative mass ratios of RA molecular components represented in the colorbar. Water content is dependent on the relative humidity of the environment and is thus omitted from the molecular ratios. Constructing the respiratory aerosolized virion model A 100 nm cubic box with the RA fluid recipe specified above was built with PACKMOL (Martínez et al., 2009), minimized, equilibrated briefly on TACC Frontera, then replicated to form a 300 nm cube. The RA box was then carved into a 270 nm diameter sphere. To make space for the placement of V within the RA, a spherical selection with volume corresponding to that of the V membrane + S crown (radius 734 Å) was deleted from the center of the RA. The final equilibrated V model, including surrounding equilibrated waters and ions (733 Å radius), was translated into the RA. Atom clashes were resolved using a 1.2 Å cutoff. Hydrogen mass repartitioning (Hopkins et al., 2015) was applied to the structure to improve performance. The simulation box was increased to 2800 Å per side to provide a 100 Å vacuum atmospheric buffer. The RAV simulation was conducted in an NVT ensemble with a 4 fs timestep. After minimizing, the RAV was heated to 298 K with 0.1 kcal/mol Å2 restraints on the viral lipid headgroups, then equilibrated for 1.5 ns. Finally, a cross-section of the RAV model-including and open S, m1/m2, and ALB (called the SMA system)-was constructed with PACKMOL to closely observe atomic scale interactions within the RAV model (Figure 4). An external file that holds a picture, illustration, etc. Object name is 10.1177_10943420221128233-fig4.jpg Figure 4. SMA system captured with multiscale modeling from classical MD to AI-enabled quantum mechanics. For all panels: S protein shown in cyan, S glycans in blue, m1/m2 shown in red, ALB in orange, Ca2+ in yellow spheres, viral membrane in purple. A) Interactions between mucins and S facilitated by glycans and Ca2+. B) Snapshot from SMA simulations. C) Example Ca2+ binding site from SMA simulations (1800 sites, each 1000+ atoms) used for AI-enabled quantum mechanical estimates from OrbNet Sky. D) Quantification of contacts between S and mucin from SMA simulations. E) OrbNet Sky energies versus CHARMM36m energies for each sub-selected system, colored by total number of atoms. Performance of OrbNet Sky versus DFT in subplot (?B97x-D3/def-TZVP, R2=0.99, for 17 systems of peptides chelating Ca2+ (Hu et al., 2021)). Visualized with VMD. Parameter evaluation with OrbNet Comparison to quantum methods reveals significant polarization effects, and shows that there is opportunity to improve the accuracy of fixed charge force fields. For the large system sizes associated with solvated Ca2+-protein interaction motifs (over 1000 atoms, even in aggressively truncated systems), conventional quantum mechanics methods like density functional theory (DFT) are impractical for analyzing a statistically significant ensemble of distinct configurations (see discussion in Performance Results). In contrast, OrbNet allows for DFT accuracy with over 1000-fold speedup, providing a useful method for benchmarking and refining the force field simulation parameters with quantum accuracy (Christensen et al., 2021). To confirm the accuracy of OrbNet versus DFT (?B97X-D/def2-TZVP), the inset of Figure 4(E) correlates the two methods for the Ca2+-binding energy in a benchmark dataset of small Ca2+-peptide complexes (Hu et al., 2021). The excellent correlation of OrbNet and DFT for the present use case is clear from the inset figure; six datapoints were removed from this plot on the basis of a diagnostic applied to the semi-empirical GFN-xTB solution used for feature generation of OrbNet (Christensen et al., 2021). Figure 4 presents a comparison of the validated OrbNet method with the CHARMM36m force field for 1800 snapshots taken from the SMA MD simulations. At each snapshot, a subsystem containing a solvated Ca2+-protein complex was extracted (Figure 4(E)), with protein bonds capped by hydrogens. For both OrbNet and the force field, the Ca2+-binding energy was computed and shown in the correlation plot. Lack of correlation between OrbNet and the force field identifies important polarization effects, absent in a fixed charge description. Similarly, the steep slope of the best-fit line in Figure 4(E) reflects the fact that some of the configurations sampled using MD with the CHARMM36m force field are relatively high in energy according to the more accurate OrbNet potential. This approach allows us to test and quantify limitations of empirical force fields, such as lack of electronic polarization. The practicality of OrbNet for these simulation snapshots with 1000+ atoms offers a straightforward multiscale strategy for refining the accuracy of the CHARMM36m force field. By optimizing the partial charges and other force field parameters, improved correlation with OrbNet for the subtle Ca2+-protein interactions could be achieved, leading to near-quantum accuracy simulations with improved configurational sampling. The calculations presented here present a proof-of-concept of this iterative strategy. AI-WE simulations of delta spike opening While our previous WE simulations of the WT SARS-CoV-2 S-opening (Sztain et al., 2021) were notable in generating pathways for a seconds-timescale process of a massive system, we have made two critical technological advancements in the WESTPA software that greatly enhance the efficiency and analysis of WE simulations. These advances enabled striking observations of Delta variant S opening (Figures 5 and and6).6). First, in contrast to prior manual bins for controlling trajectory replication, we have developed automated and adaptive binning that enables more efficient surmounting of large barriers via early identification of "bottleneck" regions (Torrillo et al., 2021). Second, we have parallelized, memory-optimized, and implemented data streaming for the history-augmented Markov state model (haMSM) analysis scheme (Copperman and Zuckerman 2020) to enable application to the TB-scale S-opening datasets. The haMSM approach estimates rate constants from simulations that have not yet reached a steady state (Suarez et al., 2014). Our WE simulations generated >800 atomically detailed, Delta variant S-opening pathways (Figures 5(B) and and6)6) of the receptor binding domain (RBD) switching from a glycan-shielded "down" to an exposed "up" state using 72 µs of total simulation time within 14 days using 192 NVIDIA V100 GPUs at a time on TACC's Longhorn supercomputer. Among these pathways, 83 reach an "open" state that aligns with the structure of the human ACE2-bound WT S protein (Benton et al., 2020) and 18 reach a dramatically open state (Figure 6). Our haMSM analysis of WT WE simulations successfully provided long-timescale (steady state) rate constants for S-opening based on highly transient information (Figure 5(C)). We also leveraged a simple, yet powerful unsupervised deep learning method called Anharmonic Conformational Analysis enabled Autoencoders (ANCA-AE) Clyde et al. (2021) to extract conformational states from our long-timescale WE simulations of Delta spike opening (Figures 5(A) and (D)). ANCA-AE first minimizes the fourth order correlations in atomistic fluctuations from MD simulation datasets and projects the data onto a low dimensional space where one can visualize the anharmonic conformational fluctuations. These projections are then input to an autoencoder that further minimizes non-linear correlations in the atomistic fluctuations to learn an embedding where conformations are automatically clustered based on their structural and energetic similarity. A visualization of the first three dimensions from the latent space articulates the RBD opening motion from its closed state (Figure 5(D)). It is notable that while other deep learning techniques need special purpose hardware (such as GPUs), the ANCA-AE approach can be run with relatively modest CPU resources and can therefore scale to much larger systems (e.g., the virion within aerosol) when optimized. D-NEMD explores pH effects on delta spike We performed D-NEMD simulations of the SH system with GROMACS (Abraham et al., 2015) using a ?pH=2.0 (from 7.0 to 5.0) as the external perturbation. We ran 3200-ns equilibrium MD simulations of SH to generate 87 configurations (29 configurations per replicate) that were used as the starting points for multiple short (10 ns) D-NEMD trajectories under the effect of the external perturbation (?pH=2.0). The effect of a ?pH was modeled by changing the protonation state of histidines 66, 69, 146, 245, 625, 655, 1064, 1083, 1088, and 1101 (we note that other residues may also become protonated (Lobo and Warwicker 2021); the D-NEMD approach can also be applied to examine those). The structural response of the S to the pH decrease was investigated by measuring the difference in the position for each Ca atom between the equilibrium and corresponding D-NEMD simulation at equivalent points in time (Oliveira et al., 2021a), namely after 0, 0.1, 1, 5, and 10 ns of simulation. The D-NEMD simulations reveal that pH changes, of the type expected in aerosols, affect the dynamics of functionally important regions of the spike, with potential implications for viral behavior (Figure 7). As this approach involves multiple short independent non-equilibrium trajectories, it is well suited for cloud computing. All D-NEMD simulations were performed using Oracle Cloud. An external file that holds a picture, illustration, etc. Object name is 10.1177_10943420221128233-fig7.jpg Figure 7. D-NEMD simulations reveal changes in key functional regions of the S protein, including the receptor binding domain, as the result of a pH decrease. Color scale and ribbon thickness indicate the degree of deviation of Ca atoms from their equilibrium position. Red spheres indicate the location of positively charged histidines. How performance was measured WESTPA For the WE simulations of spike opening using WESTPA, we defined the time-to-solution as the total simulation time required to generate the first spike opening event. Spike opening is essentially impossible to observe via conventional MD. WESTPA simulations were run using the AMBER20 dynamics engine and 192 NVIDIA V100 GPUs at a time on TACC's Longhorn supercomputer. NAMD NAMD performance metrics were collected using hardware performance counters for FLOPs/step measurements, and application-internal timers for overall simulation rates achieved by production runs including all I/O for simulation trajectory and checkpoint output. NAMD FLOPs/step measurements were conducted on TACC Frontera, by querying hardware performance counters with the rdmsr utility from Intel msr-tools1 and the "TACC stats" system programs.2 For each simulation, FLOP counts were measured for NAMD simulation runs of two different step counts. The results of the two simulation lengths were subtracted to eliminate NAMD startup operations, yielding an accurate estimate of the marginal FLOPs per step for a continuing simulation (Phillips et al., 2002). Using the FLOPs/step values computed for each simulation, overall FLOP rates were computed by dividing the FLOPs/step value by seconds/step performance data reported by NAMD internal application timers during production runs. GROMACS GROMACS 2020.4 benchmarking was performed on Oracle Cloud Infrastructure (OCI)3 compute shape BM.GPU4.8 consisting of 8×NVIDIA A100 tensor core GPUs, and 64 AMD Rome CPU cores. The simulation used for benchmarking contained 615,563 atoms and was run for 500,000 steps with 2 fs time steps. The simulations were run on increasing numbers of GPUs, from 1 to 8, using eight CPU cores per GPU, running for both the production (Nose-Hoover) and GPU-accelerated (velocity rescaling) thermostats. Particle-mesh Ewald (PME) calculations were pinned to a single GPU, with additional GPUs for multi-GPU jobs used for particle-particle calculations. Performance data (ns/day and average single-precision TFLOPS, calculated as total number of TFLOPs divided by total job walltime) were reported by GROMACS itself. Each simulation was repeated four times and average performance figures reported. Performance results Table 2. Table 2. MD simulation floating point ops per timestep. MD Simulation Code Atoms aFLOPs/step Spike, head AMBER, GROMACS 0.6 m 62.14 GFLOPs/step Spike NAMD 1.7 m 43.05 GFLOPs/step S+m1/m2+ALB NAMD 2.1 m 54.86 GFLOPs/step Resp. Aero.+Vir NAMD 1B 25.81 TFLOPs/step Open in a separate window aFLOPs/step data were computed by direct FLOP measurements from hardware performance counters for NAMD simulations, or by using the application-reported FLOP rates and ns/day simulation performance in the case of GROMACS. NAMD performance NAMD was used to perform all of the simulations listed in Table 1, except for the closed spike "SH" simulations described further below. With the exception of the aerosol and virion simulation, the other NAMD simulations used conventional protocols and have performance and parallel scaling characteristics that closely match the results reported in our previous SARS-CoV-2 research Casalino et al. (2021). NAMD 2.14 scaling performance for the one billion-atom respiratory aerosol and virion simulation run on ORNL Summit is summarized in Tables 3 and and4.4. A significant performance challenge associated with the aerosol virion simulation relates to the roughly 50% reduction in particle density as compared with a more conventional simulation with a fully populated periodic cell. The reduced particle density results in large regions of empty space that nevertheless incur additional overheads associated with both force calculations and integration, and creates problems for the standard NAMD load balancing scheme that estimates the work associated with the cubic "patches" used for parallel domain decomposition. The PME electrostatics algorithm and associated 3-D FFT and transpose operations encompass the entire simulation unit cell and associated patches, requiring involvement in communication and reduction operations despite the inclusion of empty space. Enabling NAMD diagnostic output on a 512-node 1B-atom aerosol and virion simulation revealed that ranks assigned empty regions of the periodic cell had 66 times the number of fixed-size patches as ranks assigned dense regions. The initial load estimate for an empty patch was changed from a fixed 10 atoms to a runtime parameter with a default of 40 atoms, which reduced the patch ratio from 66 to 19 and doubled performance on 512 nodes. Table 3. NAMD performance: Respiratory Aerosol + Virion, 1B atoms, 4 fs timestep w/HMR, and PME every three steps. Nodes Summit Speedup Efficiency CPU + GPU 256 4.18 ns/day ~1.0× ~100% 512 7.68 ns/day 1.84× 92% 1024 13.64 ns/day 3.27× 81% 2048 23.10 ns/day 5.53× 69% 4096 34.21 ns/day 8.19× 51% Open in a separate window Table 4. Peak NAMD FLOP rates, ORNL Summit. NAMD Simulation Atoms, B Nodes Sim rate Performance Resp. Aero.+Vir 1 4096 34.21 ns/day 2.55 PFLOPS Open in a separate window WESTPA performance Our time to solution for WE simulations of spike opening (to the "up" state) (Figure 5) using the WESTPA software and AMBER20 was 14 µs of total simulation time, which was completed in 4 days using 192 NVIDIA V100 GPUs at a time on TACC's Longhorn supercomputer. For reference, conventional MD would require an expected ~5 orders of magnitude more computing. The WESTPA software is highly scalable, with nearly perfect scaling out to >1000 NVIDIA V100 GPUs and this scaling is expected to continue until the filesystem is saturated. Thus, WESTPA makes optimal use of large supercomputers and is limited by filesystem I/O due to the periodic restarting of trajectories after short time intervals. AI-enhanced WE simulations DeepDriveMD is a framework to coordinate the concurrent execution of ensemble simulations and drive them using AI models Brace et al. (2021a); Lee et al. (2019). DeepDriveMD has been shown to improve the scientific performance of diverse problems: from-protein folding to conformation of protein-ligand complexes. We coupled WESTPA to DeepDriveMD, which is responsible for resource dynamism and concurrent heterogeneous task execution (ML and AMBER). The coupled workflow was executed on 1024 nodes on Summit (OLCF), and, in spite of the spatio-temporal heterogeneity of tasks involved, the resource utilization was in the high 90%. Consistent with earlier studies, the coupling of WESTPA to DeepDriveMD results in a 100x improvement in the exploration of phase space. GROMACS performance Figure 8 shows GROMACS parallelizes well across the eight NVIDIA A100 GPUs available on each BM.GPU4.8 instance used in the Cluster in the Cloud4 running on OCI. There is a performance drop for two GPUs due to inefficient division of the PME and particle-particle tasks. Methods to address this exist for the two GPU case Páll et al. (2020), but were not adopted as we were targeting maximum raw performance across all eight GPUs. Production simulations achieved 27% of the peak TFLOPS available from the GPUs. Multiple simulations were run across 10 such compute nodes, enabling the ensemble to run at an average combined speed of 425 TFLOPS and sampling up to 1µs/day. We note that the calculations will be able to run 20%-40% faster once the Nose-Hoover thermostat that is required for the simulation is ported to run on the GPU. Benchmarking using a velocity rescaling thermostat that has been ported to GPU shows that this would enable the simulation to extract 34% of the peak TFLOPS from the cards, enabling each node to achieve an average speed of 53.4 TFLOPS, and 125 ns/day. A cluster of 10 nodes would enable GROMACS to run at an average combined speed of over 0.5 PFLOPs, simulating over 1.2 µs/day. An external file that holds a picture, illustration, etc. Object name is 10.1177_10943420221128233-fig8.jpg Figure 8. GROMACS performance across 1-8 A100 GPUs in ns/day (thicker, blue lines) and the fraction of maximum theoretical TFLOPS (thinner, green lines); production setup shown with solid line, and runs with the GPU-accelerated thermostat in dashed. A significant innovation is that this power is available on demand: Cluster in the Cloud with GPU-optimized GROMACS was provisioned and benchmarked within 1 day of inception of the project. This was handed to the researcher, who submitted the simulations. Automatically, up to 10 BM.GPU4.8 compute nodes were provisioned on-demand based on requests from the Slurm scheduler. These simulations were performed on OCI, using Cluster in the Cloud Williams (2021) to manage automatic scaling. Cluster in the Cloud was configured to dynamically provision and terminate computing nodes based on the workload. Simulations were conducted using GROMACS 2020.4 compiled with CUDA support. Multiple simultaneous simulations were conducted, with each simulation utilizing a single BM.GPU4.8 node without multinode parallelism. This allowed all production simulations to be completed within 2 days. The actual compute cost of the project was less than $6125 USD (on-demand OCI list price). The huge reduction in "time to science" that low-cost cloud enables changes the way that researchers can access and use HPC facilities. In our opinion, such a setup enables "exclusive on-demand" HPC capabilities for the scientific community for rapid advancement in science. OrbNet performance Prior benchmarking reveals that OrbNet provides over 1000-fold speedup compared to DFT (Christensen et al., 2021). For the calculations presented here, the cost of corresponding high quality range-separated DFT calculations (?B97X-D/def2-TZVP) can be estimated. In Figure 4(E), we consider system sizes which would require 14,000-47,000 atomic orbitals for ?B97X-D/def2-TZVP, exceeding the range of typical DFT evaluations. Estimation of the DFT computational cost of the 1811 configurations studied in Figure 4(E) suggests a total of 115M core-hours on NERSC Cori Haswell nodes; in contrast, the OrbNet calculations for the current study require only 100k core-hours on the same nodes. DFT cost estimates were based on extrapolation from a dataset of over 1M ChEMBL molecules ranging in size from 40 to 107 atom systems considering only the cubic cost component of DFT (Christensen et al., 2021). Go to: Implications Our major scientific achievements are 1. We showcase an extensible AI-enabled multiscale computational framework that bridges time and length scales from electronic structure through whole aerosol particle morphology and dynamics. 2. We develop all-atom simulations of respiratory mucins, and use these to understand the structural basis of interaction with the SARS-CoV-2 spike protein. This has implications for viral binding in the deep lung, which is coated with mucins. We expect the impact of our mucin simulations to be far reaching, as malfunctions in mucin secretion and folding have been implicated in progression of severe diseases such as cancer and cystic fibrosis. 3. We present a significantly enhanced all-atom model and simulation of the SARS-CoV-2 Delta virion, which includes the hundreds of tiled M-protein dimers and the E-protein ion channels. This model can be used as a basis to understand why the Delta virus is so much more infectious than the WT or alpha variants. 4. We develop an ultra-large (1 billion+) all-atom simulation capturing massive chemical and biological complexity within a respiratory aerosol. This simulation provides the first atomic-level views of virus-laden aerosols and is already serving as a basis to develop an untold number of experimentally testable hypotheses. An immediate example suggests a mechanism through which mucins and other species, for example, lipids, which are present in the aerosol, arrange to protect the molecular structure of the virus, which otherwise would be exposed to the air-water interface. This work also opens the door for developing simulations of other aerosols, for example, sea spray aerosols, that are involved in regulating climate. 5. We evidence how changes in pH, which are expected in the aerosol environment, may alter dynamics and allosteric communication pathways in key functional regions of the Delta spike protein. 6. We characterize atomically detailed pathways for the spike-opening process of the Delta variant using WE simulations, revealing a dramatically open state that may facilitate binding to human host cells. 7. We demonstrate how parallelized haMSM analysis of WE data can provide physical rate estimates of spike opening, improving prior estimates by many orders of magnitude. The pipeline can readily be applied to the any variant spike protein or other complex systems of interest. 8. We show how HPC and cloud resources can be used to significantly drive down time-to-solution for major scientific efforts as well as connect researchers and greatly enable complex collaborative interactions. 9. We demonstrate how AI coupled to HPC at multiple levels can result in significantly improved effective performance, for example, with AI-driven WESTPA, and extend the reach and domain of applicability of tools ordinarily restricted to smaller, less complex systems, for example, with OrbNet. 10. While our work provides a successful use case, it also exposes weaknesses in the HPC ecosystem in terms of support for key steps in large/complex computational science campaigns. We find lack of widespread support for high performance remote visualization and interactive graphical sessions for system preparation, debugging, and analysis with diverse science tools to be a limiting factor in such efforts. Go to: Acknowledgements We thank Prof. Kim Prather for inspiring and informative discussions about aerosols and for her commitment to convey the airborne nature of SARS-CoV-2. We thank D. Veesler for sharing the Delta spike NTD coordinates in advance of publication. We thank B. Messer, D. Maxwell, and the Oak Ridge Leadership Computing Facility at Oak Ridge National Laboratory supported by the DOE under Contract DE-AC05-00OR22725. We thank the Texas Advanced Computing Center Frontera team, especially D. Stanzione and T. Cockerill, and for compute time made available through a Director's Discretionary Allocation (NSF OAC-1818253). We thank the Argonne Leadership Computing Facility supported by the DOE under DE-AC02-06CH11357. We thank the Pittsburgh Supercomputer Center for providing priority queues on Bridges-2 through the XSEDE allocation NSF TG-CHE060063. We thank N. Kern and J. Lee of the CHARMM-GUI support team for help converting topologies between NAMD and GROMACS. We thank J. Copperman, G. Simpson, D. Aristoff, and J. Leung for valuable discussions and support from NIH grant GM115805. NAMD and VMD are funded by NIH P41-GM104601. This work was supported by the NSF Center for Aerosol Impacts on Chemistry of the Environment (CAICE), National Science Foundation Center for Chemical Innovation (NSF CHE-1801971), as well as NIH GM132826, NSF RAPID MCB-2032054, an award from the RCSA Research Corp., a UC San Diego Moore's Cancer Center 2020 SARS-CoV-2 seed grant, to R.E.A. This work was also supported by Oracle Cloud credits and related resources provided by the Oracle for Research program. AJM and ASFO receive funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (PREDACTED Advanced Grant, Grant agreement No.: 101021207). Go to: Notes 1. https://github.com/intel/msr-tools 2. https://github.com/TACC/tacc_stats 3. https://www.oracle.com/cloud/ 4. https://cluster-in-the-cloud.readthedocs.io/ Go to: Footnotes The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by National Science Foundation (CHE- 1801971); National Science Foundation (MCB- 2032054); National Science Foundation (OAC-1818253); National Science Foundation (TG-CHE060063); U.S. Department of Energy (DE-AC02-06CH11357); U.S. Department of Energy (DE-AC05- 00OR22725); National Institutes of Health (P41-GM104601); National Institutes of Health (R01-GM132826). Go to: ORCID iDs Lorenzo Casalino https://orcid.org/0000-0003-3581-1148 Mia Rosenfeld https://orcid.org/0000-0002-8961-8231 Surl-Hee Ahn https://orcid.org/0000-0002-3422-805X John Russo https://orcid.org/0000-0002-2813-6554 Sofia Oliveira https://orcid.org/0000-0001-8753-4950 Clare Morris https://orcid.org/0000-0002-4314-5387 Alexander Brace https://orcid.org/0000-0001-9873-9177 Hyungro Lee https://orcid.org/0000-0002-4221-7094 Zhuoran Qiao https://orcid.org/0000-0002-5704-7331 Anima Anandkumar https://orcid.org/0000-0002-6974-6797 James Phillips https://orcid.org/0000-0002-2296-3591 John McCalpin https://orcid.org/0000-0002-2535-1355 Christopher Woods https://orcid.org/0000-0001-6563-9903 Matt Williams https://orcid.org/0000-0003-2198-1058 Richard Pitts https://orcid.org/0000-0002-2037-3360 Daniel Zuckerman https://orcid.org/0000-0001-7662-2031 Adrian Mulholland https://orcid.org/0000-0003-1015-4567 Arvind Ramanathan https://orcid.org/0000-0002-1622-5488 Lillian Chong https://orcid.org/0000-0002-0590-483X Rommie E Amaro https://orcid.org/0000-0002-9275-9553 Go to: References Abraham MJ, Murtola T, Schulz R, et al. (2015) GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1-2: 19-25. DOI: 10.1016/j.softx.2015.06.001 10.1016/j.softx.2015.06.001. [CrossRef] [CrossRef] [Google Scholar] Adhikari U, Mostofian B, Copperman J, et al. (2019) Computational estimation of ms-sec atomistic folding times. Journal of the American Chemical Society 141: 6519-6526. DOI: 10.1101/427393 10.1101/427393. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Bangaru S, Gabriel O, Turner HL, et al. (2020) Structural analysis of full-length SARS-CoV-2 spike protein from an advanced vaccine candidate. Science 370: 65201089-65201094. DOI: 10.1126/science.abe1502 10.1126/science.abe1502. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Beglov D, Roux B. (1994) Finite representation of an infinite bulk system: Solvent boundary potential for computer simulations. The Journal of Chemical Physics 100(12). DOI: 10.1063/1.466711 10.1063/1.466711. [CrossRef] [CrossRef] [Google Scholar] Benton DJ, Wrobel AG, Xu P, et al. (2020) Receptor binding and priming of the spike protein of SARS-CoV-2 for membrane fusion. Nature 588: 7837327-7837330. DOI: 10.1038/s41586-020-2772-0 10.1038/s41586-020-2772-0. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Bhowmik D, Gao S, Young MT, et al. (2018) Deep clustering of protein folding simulations. BMC Bioinformatics 19(18): 484. DOI: 10.1186/s12859-018-2507-5 10.1186/s12859-018-2507-5. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Blender Online Community (2020) Blender - a 3D modelling and rendering package. http://www.blender.org. Brace A, Lee H, Ma H, et al. (2021. a) Achieving 100X Faster Simulations of Complex Biological Phenomena by Coupling ML to HPC Ensembles. arXiv: cs.DC/2104.04797. [Google Scholar] Brace A, Michael S, Subbiah V, et al. (2021. b) Stream-AI-MD: Streaming AI-Driven Adaptive Molecular Simulations for Heterogeneous Computing Platforms. New York, NY, USA: Association for Computing Machinery. DOI: 10.1145/3468267.3470578 10.1145/3468267.3470578. [CrossRef] [CrossRef] [Google Scholar] Casalino L, C Dommer A, Gaieb Z, et al. (2021) AI-driven multiscale simulations illuminate mechanisms of SARS-CoV-2 spike dynamics. The International Journal of High Performance Computing Applications 35(5): 432-451. DOI: 10.1177/10943420211006452 10.1177/10943420211006452. [CrossRef] [CrossRef] [Google Scholar] Casalino L, Gaieb Z, Goldsmith JA, et al. (2020) Beyond Shielding: The Roles of Glycans in the SARS-CoV-2 Spike Protein. ACS Central Science 6(10): 1722-1734. DOI: 10.1021/acscentsci.0c01056 10.1021/acscentsci.0c01056. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Case DA, Cheatham TE, III, Darden TA, et al. (n.d.). San Francisco: Publisher: University of California. Amber16. ([n. d.]). [Google Scholar] Chai J, Cai Y, Pang C, et al. (2021) Structural basis for SARS-CoV-2 envelope protein recognition of human cell junction protein PALS1. Nature Communications 12(1): 3433. DOI: 10.1038/s41467-021-23533-x 10.1038/s41467-021-23533-x. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Christensen AS, Krishna Sirumalla S, Qiao Z, et al. (2021) OrbNet Denali: A Machine Learning Potential for Biological and Organic Chemistry with Semi-empirical Cost and DFT Accuracy. arXiv: physics.chem-ph/2107.00299. [PubMed] [Google Scholar] Ciccotti G, Ferrario M. (2016) Non-equilibrium by molecular dynamics: a dynamical approach. Molecular Simulation 42(16): 1385-1400. DOI: 10.1080/08927022.2015.1121543 10.1080/08927022.2015.1121543. [CrossRef] [CrossRef] [Google Scholar] Clyde A, Galanie S, Kneller DW, et al. (2021) High Throughput Virtual Screening and Validation of a SARS-CoV-2 Main Protease Non-covalent Inhibitor. bioRxiv. arXiv: DOI: 10.1101/2021.03.27.437323. https://www.biorxiv.org/content/early/2021/04/02/2021.03.27.437323.full.pdf [PMC free article] [PubMed] [CrossRef] [Google Scholar] Coleman KK, Wen Tay DJ, Tan KS, et al. (2021) Viral Load of SARS-CoV-2 in Respiratory Aerosols Emitted by COVID-19 Patients while Breathing, Talking, and Singing. Clinical Infectious Diseases. DOI: 10.1093/cid/ciab691 10.1093/cid/ciab691. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Copperman J, Zuckerman DM. (2020) Accelerated estimation of long-timescale kinetics from weighted ensemble simulation via non-Markovian "microbin" analysis. Journal of Chemical Theory and Computation 16(11): 6763-6775. [PMC free article] [PubMed] [Google Scholar] D'Imprima E, Floris D, Joppe M, et al. (2019) Protein denaturation at the air-water interface and how to prevent it. eLife 8: e42747. DOI: 10.7554/eLife.42747 10.7554/eLife.42747. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Durrant JD, Amaro RE. (2014) LipidWrapper: An Algorithm for Generating Large-Scale Membrane Models of Arbitrary Geometry. PLoS Computational Biology 10: 7. DOI: 10.1371/journal.pcbi.1003720 10.1371/journal.pcbi.1003720. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Fennelly KP. (2020) Particle sizes of infectious aerosols: implications for infection control. The Lancet Respiratory Medicine 8(9): 914-924. DOI: 10.1016/S2213-2600(20)30323-4 10.1016/S2213-2600(20)30323-4. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Galdadas I, Shen Q, F Oliveira AS, et al. (2021) Allosteric communication in class A ß-lactamases occurs via cooperative coupling of loop dynamics. eLife 10: e66567. DOI: 10.7554/eLife.66567 10.7554/eLife.66567. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Guvench O, Hatcher E, Venable RM, et al. (2009) CHARMM additive all-atom force field for glycosidic linkages between hexopyranoses. Journal of Chemical Theory and Computation 5: 2353-2370. DOI: 10.1021/ct900242e 10.1021/ct900242e. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Han K, Richard M, VenableBryant A-M, et al. (2018) Graph-Theoretic Analysis of Monomethyl Phosphate Clustering in Ionic Solutions. The Journal of Physical Chemistry B 122(4): 1484-1494. PMID: 29293344 DOI: 10.1021/acs.jpcb.7b10730 10.1021/acs.jpcb.7b10730. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Hopkins CW, ScottGrand Le, Walker RC, et al. (2015) Long-Time-Step Molecular Dynamics through Hydrogen Mass Repartitioning. Journal of Chemical Theory and Computation 11(1): 1864-1874. DOI: 10.1021/ct5010406 10.1021/ct5010406. [PubMed] [CrossRef] [CrossRef] [Google Scholar] Hu X, Lenz-Himmer M-O, Baldauf C. (2021) Better Force Fields Start with Better Data - A Data Set of Cation Dipeptide Interactions. arXiv:q-bio.BM/2107.08855. [PMC free article] [PubMed] [Google Scholar] Huang J, Mackerell AD. (2013) CHARMM36 all-atom additive protein force field: Validation based on comparison to NMR data. Journal of Computational Chemistry 34(25): 2135-2145. DOI: 10.1002/jcc.23354 10.1002/jcc.23354. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Huang J, Rauscher S, Nawrocki G, et al. (2017) CHARMM36m: An Improved Force Field for Folded and Intrinsically Disordered Proteins. Nature Methods 14(1): 71-73. DOI: 10.1038/nmeth.4067 10.1038/nmeth.4067. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Huber GA, Kim S. (1996) Weighted-ensemble Brownian dynamics simulations for protein association reactions. Biophysical Journal 70(1): 97-110. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1224912/ [PMC free article] [PubMed] [Google Scholar] Hughes GW, Ridley C, Collins R, et al. (2019) The MUC5B mucin polymer is dominated by repeating structural motifs and its topology is regulated by calcium and pH. Scientific Reports 9(1): 17350. DOI: 10.1038/s41598-019-53768-0 10.1038/s41598-019-53768-0. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Humphrey W, Dalke A, Schulten K. (1996) VMD - Visual Molecular Dynamics. J. Mol. Graphics 14(1): 33-38. DOI: 10.1016/0263-7855(96)00018-5 10.1016/0263-7855(96)00018-5. [PubMed] [CrossRef] [CrossRef] [Google Scholar] Jo S, Song KC, Desaire H, et al. (2011) Glycan reader: Automated sugar identification and simulation preparation for carbohydrates and glycoproteins. Journal of Computational Chemistry 32(14): 3135-3141. DOI: 10.1002/jcc.21886 10.1002/jcc.21886. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Kannan SR, Spratt AN, Cohen AR, et al. (2021) Evolutionary analysis of the Delta and Delta Plus variants of the SARS-CoV-2 viruses. Journal of Autoimmunity 124(2021): 102715. DOI: 10.1016/j.jaut.2021.102715 10.1016/j.jaut.2021.102715. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Ke Z, Oton J, Qu K, et al. (2020) Structures and distributions of SARS-CoV-2 spike proteins on intact virions. Nature 588(2020): 1-7. DOI: 10.1038/s41586-020-2665-2 10.1038/s41586-020-2665-2. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Klauda JB, Venable RM, Alfredo Freites J, et al. (2010) Update of the CHARMM All-Atom Additive Force Field for Lipids: Validation on Six Lipid Types. The Journal of Physical Chemistry B 114(23): 7830-7843. PMID: 20496934 DOI: 10.1021/jp101759q 10.1021/jp101759q. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Lee H, Turilli M, Jha S, et al. (2019) DeepDriveMD: Deep-Learning Driven Adaptive Molecular Simulations for Protein Folding.In: 2019 IEEE/ACM Third Workshop on Deep Learning on Supercomputers (DLS), pp. 12-19. [Google Scholar] Lee J, Cheng Xi, Jason M, et al. (2016) CHARMM-GUI Input Generator for NAMD, GROMACS, AMBER, OpenMM, and CHARMM/OpenMM Simulations Using the CHARMM36 Additive Force Field. Journal of Chemical Theory and Computation 12(1): 405-413. PMID: 26631602 DOI: 10.1021/acs.jctc.5b00935 10.1021/acs.jctc.5b00935. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Lobo VR, Warwicker J. (2021) Predicted pH-dependent stability of SARS-CoV-2 spike protein trimer from interfacial acidic groups. Computational and Structural Biotechnology Journal 19(2021): 5140-5148. DOI: 10.1016/j.csbj.2021.08.049 10.1016/j.csbj.2021.08.049. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Mandala VS, McKay MJ, Shcherbakov AA, et al. (2020) Structure and drug binding of the SARS-CoV-2 envelope protein transmembrane domain in lipid bilayers. Nature Structural & Molecular Biology 27(12): 1202-1208. DOI: 10.1038/s41594-020-00536-8 10.1038/s41594-020-00536-8. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Mariethoz J, Alocci D, Gastaldello A, et al. (2018) Glycomics@ExPASy: Bridging the Gap*. Molecular and Cellular Proteomics 17(11): 2164-2176. DOI: 10.1074/mcp.RA118.000799 10.1074/mcp.RA118.000799. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Markovetz MR, Subramani DB, Kissner WJ, et al. (2019) Endotracheal tube mucus as a source of airway mucus for rheological study. American Journal of Physiology-Lung Cellular and Molecular Physiology 317(4): L498-L509. PMID: 31389736 DOI: 10.1152/ajplung.00238.2019 10.1152/ajplung.00238.2019. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Martínez L, Andrade R, Birgin EG, et al. (2009) PACKMOL: A package for building initial configurations for molecular dynamics simulations. Journal of Computational Chemistry 30: 132157-132164. arXiv: DOI: 10.1002/jcc.21224. [PubMed] [CrossRef] [Google Scholar] McCallum M, Walls AC, Sprouse KR, et al. (2021) Molecular Basis of Immune Evasion by the Delta and Kappa SARS-CoV-2 Variants. bioRxiv. arXiv: DOI: 10.1101/2021.08.11.455956. https://www.biorxiv.org/content/early/2021/08/12/2021.08.11.455956.full.pdf [PubMed] [CrossRef] [Google Scholar] Miller SL, Nazaroff WW, Jimenez JL, et al. (2021) Transmission of SARS-CoV-2 by inhalation of respiratory aerosol in the Skagit Valley Chorale superspreading event. Indoor Air 31(2): 314-323. DOI: 10.1111/ina.12751 10.1111/ina.12751. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Noe F. (2020) Machine Learning for Molecular Dynamics on Long Timescales. Cham: Springer International Publishing, pp. 331-372. DOI: 10.1007/978-3-030-40245-7_16 10.1007/978-3-030-40245-7_16. [CrossRef] [CrossRef] [Google Scholar] Oliveira ASF, Ciccotti G, Haider S, et al. (2021. a) Dynamical nonequilibrium molecular dynamics reveals the structural basis for allostery and signal propagation in biomolecular systems. The European Physical Journal B 94(7): 144. DOI: 10.1140/epjb/s10051-021-00157-0 10.1140/epjb/s10051-021-00157-0. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Oliveira ASF, Edsall CJ, Woods CJ, et al. (2019) A General Mechanism for Signal Propagation in the Nicotinic Acetylcholine Receptor Family. Journal of the American Chemical Society 141(51): 19953-19958. PMID: 31805762 DOI: 10.1021/jacs.9b09055 10.1021/jacs.9b09055. [PubMed] [CrossRef] [CrossRef] [Google Scholar] Oliveira ASF, Shoemark DK, Avila Ibarra A, et al. (2021. b) The fatty acid site is coupled to functional motifs in the SARS-CoV-2 spike protein and modulates spike allosteric behaviour. bioRxiv 20arXiv: DOI: 10.1101/2021.06.07.447341. https://www.biorxiv.org/content/early/2021/06/09/2021.06.07.447341.full.pdf [PMC free article] [PubMed] [CrossRef] [Google Scholar] Park SJ, Lee J, Qi Y, et al. (2019) CHARMM-GUI Glycan Modeler for modeling and simulation of carbohydrates and glycoconjugates. Glycobiology 29(4): 320-331. DOI: 10.1093/glycob/cwz003 10.1093/glycob/cwz003. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Petters MD, Kreidenweis SM. (2007) A single parameter representation of hygroscopic growth and cloud condensation nucleus activity. Atmospheric Chemistry and Physics 7(8): 1961-1971. DOI: 10.5194/acp-7-1961-2007 10.5194/acp-7-1961-2007. [CrossRef] [CrossRef] [Google Scholar] Phillips J, Zheng G, Kumar S, et al. (2002) NAMD: Biomolecular Simulation on Thousands of Processors. In:Proceedings of the IEEE/ACM SC2002 Conference. Baltimore, MD: IEEE Press, pp. 1-18. Technical Paper 277. [Google Scholar] Phillips J., Braun R, Wang W, et al. (2005) Scalable Molecular Dynamics with NAMD. DOI: 10.1002/jcc.20289 10.1002/jcc.20289. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Phillips JC, Hardy DJ, Maia JDC, et al. (2020) Scalable molecular dynamics on CPU and GPU architectures with NAMD. J. Chem. Phys 153: 044130. DOI: 10.1063/5.0014475 10.1063/5.0014475. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Páll S, Zhmurov A, Bauer P, et al. (2020) Heterogeneous parallelization and acceleration of molecular dynamics simulations in GROMACS. The Journal of Chemical Physics 153: 13134110. DOI: 10.1063/5.0018516 10.1063/5.0018516. [PubMed] [CrossRef] [CrossRef] [Google Scholar] Qiao Z, Welborn M, Anandkumar A, et al. (2020) OrbNet: Deep learning for quantum chemistry using symmetry-adapted atomic-orbital features. The Journal of Chemical Physics 153: 12124111. DOI: 10.1063/5.0021955 10.1063/5.0021955. [PubMed] [CrossRef] [CrossRef] [Google Scholar] Raveh B, London N, Schueler-Furman O. (2010) Sub-angstrom modeling of complexes between flexible peptides and globular proteins. Proteins: Structure, Function, and Bioinformatics 78(9): 2029-2040. arXiv: DOI: 10.1002/prot.22716. https://onlinelibrary.wiley.com/doi/pdf/10.1002/prot.22716 [PubMed] [CrossRef] [Google Scholar] Russo JD, Zhang S, Leung JMG, et al. (2022) WESTPA 2.0: High-Performance Upgrades for Weighted Ensemble Simulations and Analysis of Longer-Timescale Applications. Journal of Chemical Theory and Computation 18: 638-649. DOI: 10.1021/acs.jctc.1c01154. [PMC free article] [PubMed] [CrossRef] [Google Scholar] Saglam AS, Chong LT. (2019) Protein-protein binding pathways and calculations of rate constants using fully-continuous, explicit-solvent simulations. Chemical Science 10(8): 2360-2372. DOI: 10.1039/c8sc04811h 10.1039/c8sc04811h. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Salomon-Ferrer R, Götz AW, Duncan P, et al. (2013) Routine Microsecond Molecular Dynamics Simulations with AMBER on GPUs. 2. Explicit Solvent Particle Mesh Ewald. Journal of Chemical Theory and Computation 9(9): 3878-3888. DOI: 10.1021/ct400314y 10.1021/ct400314y. [PubMed] [CrossRef] [CrossRef] [Google Scholar] Sener Melih, Levy Stuart, Stone John E., et al. (2021) Multiscale Modeling and Cinematic Visualization of Photosynthetic Energy Conversion Processes from Electronic to Cell Scales. Parallel Comput., p. 102698. [PMC free article] [PubMed] [Google Scholar] Stone JE, Hynninen A-P, Phillips JC, et al. (2016. a) Early Experiences Porting the NAMD and VMD Molecular Simulation and Analysis Software to GPU-Accelerated OpenPOWER Platforms. International Workshop on OpenPOWER for HPC, pp. 188-206. (IWOPH'16). [PMC free article] [PubMed] [Google Scholar] Stone JE, Barry I, Schulten K. (2013. a) Early Experiences Scaling VMD Molecular Visualization and Analysis Jobs on Blue Waters. In: Extreme Scaling Workshop (XSW), pp. 43-50. DOI: 10.1109/XSW.2013.10 10.1109/XSW.2013.10. [CrossRef] [CrossRef] [Google Scholar] Stone JE, Sener M, Vandivort KL, et al. (2016. b) Atomic Detail Visualization of Photosynthetic Membranes with GPU-Accelerated Ray Tracing. Parallel Comput 55: 17-27. DOI: 10.1016/j.parco.2015.10.015 10.1016/j.parco.2015.10.015. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Stone JE, Vandivort KL, Schulten K. (2013. b) GPU-Accelerated Molecular Visualization on Petascale Supercomputing Platforms. In: Proceedings of the 8th International Workshop on Ultrascale Visualization (UltraVis '13). New York, NY, USA: ACM, p. 8. Article 6. [Google Scholar] Suarez E, Lettieri S, Zwier MC, et al. (2014) Simultaneous computation of dynamical and equilibrium information using a weighted ensemble of trajectories. Journal of Chemical Theory and Computation 10(7): 2658-2667. [PMC free article] [PubMed] [Google Scholar] Sugio S, Kashima A, Mochizuki S, et al. (1999) Crystal structure of human serum albumin at 2.5 Å resolution. Protein Engineering, Design and Selection 12(6): 439-446. arXiv: DOI: 10.1093/protein/12.6.439 10.1093/protein/12.6.439. https://academic.oup.com/peds/article-pdf/12/6/439/18543407/120439.pdf [PubMed] [CrossRef] [CrossRef] [Google Scholar] Surya W, Li Y, Torres J. (2018) Structural model of the SARS coronavirus E channel in LMPG micelles. Biochimica et Biophysica Acta (BBA) - Biomembranes 1860(6): 1309-1317. DOI: 10.1016/j.bbamem.2018.02.017 10.1016/j.bbamem.2018.02.017. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Symmes BA, Stefanski AL, Magin CM, et al. (2018) Role of mucins in lung homeostasis: regulated expression and biosynthesis in health and disease. Biochemical Society Transactions 46(3): 707-719. arXiv: DOI: 10.1042/BST20170455 10.1042/BST20170455. https://portlandpress.com/biochemsoctrans/article-pdf/46/3/707/479418/bst-2017-0455c.pdf [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Sztain T, Ahn S-H, Bogetti AT, et al. (2021) A glycan gate controls opening of the SARS-CoV-2 spike protein. Nature Chemistry 13(10): 963-968. DOI: 10.1038/s41557-021-00758-3 10.1038/s41557-021-00758-3. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Thomsson KA, Schulz BL, Packer NH, et al. (2005) MUC5B glycosylation in human saliva reflects blood group and secretor status. Glycobiology 15(8): 791-804. arXiv: DOI: 10.1093/glycob/cwi059. https://academic.oup.com/glycob/article-pdf/15/8/791/1787060/cwi059.pdf [PubMed] [CrossRef] [Google Scholar] Torrillo PA, Bogetti AT, Chong LT. (2021) A Minimal, Adaptive Binning Scheme for Weighted Ensemble Simulations. The Journal of Physical Chemistry A 125(7): 1642-1649. DOI: 10.1021/acs.jpca.0c10724 10.1021/acs.jpca.0c10724. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Turonova B, Sikora M, Schürmann C, et al. (2020) In situ structural analysis of SARS-CoV-2 spike reveals flexibility mediated by three hinges. Science 370(6513): 203-208. DOI: 10.1126/science.abd5223 10.1126/science.abd5223. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Vejerano EP, Marr LC. (2018) Physico-chemical characteristics of evaporating respiratory fluid droplets. Journal of the Royal Society Interface 15(139): 1-10. DOI: 10.1098/rsif.2017.0939 10.1098/rsif.2017.0939. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Venable RM, Luo Y, Gawrisch K, et al. (2013) Simulations of Anionic Lipid Membranes: Development of Interaction-Specific Ion Parameters and Validation Using NMR Data. The Journal of Physical Chemistry B 117(35): 10183-10192. PMID: 23924441 DOI: 10.1021/jp401512z 10.1021/jp401512z. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Walker JS, Archer J, Florence K, et al. (2021) Accurate Representations of the Microphysical Processes Occurring during the Transport of Exhaled Aerosols and Droplets. ACS Central Science 7(1). DOI: 10.1021/acscentsci.0c01522 10.1021/acscentsci.0c01522. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Walls AC, YoungPark J, Alejandra Tortorici M, et al. (2020) Structure, Function, and Antigenicity of the SARS-CoV-2 Spike Glycoprotein. Cell 181(2): 281-292. DOI: 10.1016/j.cell.2020.02.058 10.1016/j.cell.2020.02.058. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Wang CC, Prather KA, Sznitman J, et al. (2021) Airborne transmission of respiratory viruses. Science 373(6558): eabd9149. DOI: 10.1126/science.abd9149 10.1126/science.abd9149. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Warwicker J. (2021) A model for pH coupling of the SARS-CoV-2 spike protein open/closed equilibrium. Briefings in Bioinformatics 22(2): 1499-1507. arXiv: 10.1093/bib/bbab056. https://academic.oup.com/bib/article-pdf/22/2/1499/36654668/bbab056.pdf [PMC free article] [PubMed] [CrossRef] [Google Scholar] Williams M. (2021) Cluster in the Cloud. https://cluster-in-the-cloud.readthedocs.io [Google Scholar] Wrapp D, Wang N, Corbett KS, et al. (2020) Cryo-EM structure of the 2019-nCoV spike in the prefusion conformation. Science 367(6483): 1260-1263. DOI: 10.1126/science.abb2507 10.1126/science.abb2507. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Zimmerman MI, Porter JR, Ward MD, et al. (2021) SARS-CoV-2 simulations go exascale to predict dramatic spike opening and cryptic pockets across the proteome. Nature Chemistry 13(7): 651-659. DOI: 10.1038/s41557-021-00707-0 10.1038/s41557-021-00707-0. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Zuckerman DM, Chong LT. (2017) Weighted Ensemble Simulation: Review of Methodology, Applications, and Software. Annual Review of Biophysics 46: 43-57. DOI: 10.1146/annurev-biophys-070816-033834 10.1146/annurev-biophys-070816-033834. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] Zwier MC, Adelman JL, Kaus JW, et al. (2015) WESTPA: An Interoperable, Highly Scalable Software Package for Weighted Ensemble Simulation and Analysis. Journal of Chemical Theory and Computation 11(2): 800-809. DOI: 10.1021/ct5010615 10.1021/ct5010615. [PMC free article] [PubMed] [CrossRef] [CrossRef] [Google Scholar] OTHER FORMATS PubReader PDF (2.8M) ACTIONS Cite Collections SHARE RESOURCES Similar articles Cited by other articles Links to NCBI Databases FOLLOW NCBI Connect with NLM National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Tell us what you think!Close |
| Title | A Multiscale Workflow for Modelling Ligand Complexes of Zinc Metalloproteins |
| Description | Representative MD trajectories, topologies and input files for the protein:ligand complexes presented in the paper Yang et al. J Chem. Inf. Model. 2021. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2021 |
| Provided To Others? | Yes |
| Impact | Zinc metalloproteins are ubiquitous, with protein zinc centers of structural and functional importance, involved in interactions with ligands and substrates and often of pharmacological interest. Biomolecular simulations are increasingly prominent in investigations of protein structure, dynamics, ligand interactions, and catalysis, but zinc poses a particular challenge, in part because of its versatile, flexible coordination. A computational workflow generating reliable models of ligand complexes of biological zinc centers would find broad application. Here, we evaluate the ability of alternative treatments, using (nonbonded) molecular mechanics (MM) and quantum mechanics/molecular mechanics (QM/MM) at semiempirical (DFTB3) and density functional theory (DFT) levels of theory, to describe the zinc centers of ligand complexes of six metalloenzyme systems differing in coordination geometries, zinc stoichiometries (mono- and dinuclear), and the nature of interacting groups (specifically the presence of zinc-sulfur interactions). MM molecular dynamics (MD) simulations can overfavor octahedral geometries, introducing additional water molecules to the zinc coordination shell, but this can be rectified by subsequent semiempirical (DFTB3) QM/MM MD simulations. B3LYP/MM geometry optimization further improved the accuracy of the description of coordination distances, with the overall effectiveness of the approach depending upon factors, including the presence of zinc-sulfur interactions that are less well described by semiempirical methods. We describe a workflow comprising QM/MM MD using DFTB3 followed by QM/MM geometry optimization using DFT (e.g., B3LYP) that well describes our set of zinc metalloenzyme complexes and is likely to be suitable for creating accurate models of zinc protein complexes when structural information is more limited. |
| URL | https://data.bris.ac.uk/data/dataset/10p78zgsappbz226bzrdagabq9/ |
| Title | Crystallography and QM/MM Simulations Identify Preferential Binding of Hydrolyzed Carbapenem and Penem Antibiotics to the L1 Metallo-beta-lactamase in the Imine Form |
| Description | MD trajectories, topologies, parameters and input files for the data presented in the paper Twidale et al. J Chem. Inf. Model. 2021. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2021 |
| Provided To Others? | Yes |
| Impact | Widespread bacterial resistance to carbapenem antibiotics is an increasing global health concern. Resistance has emerged due to carbapenem-hydrolyzing enzymes, including metallo-ß-lactamases (MßLs), but despite their prevalence and clinical importance, MßL mechanisms are still not fully understood. Carbapenem hydrolysis by MßLs can yield alternative product tautomers with the potential to access different binding modes. Here, we show that a combined approach employing crystallography and quantum mechanics/molecular mechanics (QM/MM) simulations allow tautomer assignment in MßL:hydrolyzed antibiotic complexes. Molecular simulations also examine (meta)stable species of alternative protonation and tautomeric states, providing mechanistic insights into ß-lactam hydrolysis. We report the crystal structure of the hydrolyzed carbapenem ertapenem bound to the L1 MßL from Stenotrophomonas maltophilia and model alternative tautomeric and protonation states of both hydrolyzed ertapenem and faropenem (a related penem antibiotic), which display different binding modes with L1. We show how the structures of both complexed ß-lactams are best described as the (2S)-imine tautomer with the carboxylate formed after ß-lactam ring cleavage deprotonated. Simulations show that enamine tautomer complexes are significantly less stable (e.g., showing partial loss of interactions with the L1 binuclear zinc center) and not consistent with experimental data. Strong interactions of Tyr32 and one zinc ion (Zn1) with ertapenem prevent a C6 group rotation, explaining the different binding modes of the two ß-lactams. Our findings establish the relative stability of different hydrolyzed (carba)penem forms in the L1 active site and identify interactions important to stable complex formation, information that should assist inhibitor design for this important antibiotic resistance determinant. |
| URL | https://data.bris.ac.uk/data/dataset/13pu85dfaobij2rumzql5buyy2/ |
| Description | Biomolecular simulation for TB drug lead discovery |
| Organisation | Ubon Ratchathani University |
| Country | Thailand |
| Sector | Academic/University |
| PI Contribution | The enoyl-acyl carrier protein reductase InhA of Mycobacterium tuberculosis is an attractive, validated target for antituberculosis drug development. Moreover, direct inhibitors of InhA remain effective against InhA variants with mutations associated with isoniazid resistance, offering the potential for activity against MDR isolates. Here, structure-based virtual screening supported by biological assays was applied to identify novel InhA inhibitors as potential antituberculosis agents. High-speed Glide SP docking was initially performed against two conformations of InhA differing in the orientation of the active site Tyr158. The resulting hits were filtered for drug-likeness based on Lipinski's rule and avoidance of PAINS-like properties and finally subjected to Glide XP docking to improve accuracy. Sixteen compounds were identified and selected for in vitro biological assays, of which two (compounds 1 and 7) showed MIC of 12.5 and 25 µg/mL against M. tuberculosis H37Rv, respectively. Inhibition assays against purified recombinant InhA determined IC50 values for these compounds of 0.38 and 0.22 µM, respectively. A crystal structure of the most potent compound, compound 7, bound to InhA revealed the inhibitor to occupy a hydrophobic pocket implicated in binding the aliphatic portions of InhA substrates but distant from the NADH cofactor, i.e., in a site distinct from those occupied by the great majority of known InhA inhibitors. This compound provides an attractive starting template for ligand optimization aimed at discovery of new and effective compounds against M. tuberculosis that act by targeting InhA. |
| Collaborator Contribution | Discovery of new and potent InhA inhibitors as antituberculosis agents: Structure-based virtual screening validated by biological assays and X-ray crystallography. Tuberculosis (TB) caused by Mycobacterium tuberculosis (M. tuberculosis) remains a major worldwide public health problem, especially in areas of high population density and low- and middle-income countries. It is the leading cause of death by infectious disease and the ninth leading overall cause of death worldwide. World Health Organization (WHO) data identified 1.6 million TB deaths and 10 million new TB cases in 2017. (1) Although TB is considered treatable, this is threatened by the spread of drug-resistant strains; it is estimated that globally there are 4.9 million cases of patients infected with multidrug-resistant tuberculosis (MDR-TB) strains resistant to isoniazid and rifampicin, the two most important anti-TB agents. In 2017, 558 000 new cases of TB were identified that were resistant to rifampicin (RR-TB), the most effective first-line drug, with 82% of these MDR-TB. About 8% of TB patients worldwide are estimated to be infected with rifampicin-susceptible, isoniazid-resistant strains (HR-TB). (2) The M. tuberculosis enoyl-acyl carrier protein (ACP) reductase (M. tuberculosis InhA) is an attractive potential target for development of new antituberculosis drugs. InhA catalyzes the NADH-specific reduction of 2-trans-enoyl-ACP (Figure 1A) in the elongation cycle of the fatty acid synthase type II (FAS II) pathway, the final step of fatty acid biosynthesis in M. tuberculosis. (3,4) InhA is the primary target of isoniazid (INH), the second first-line drug for tuberculosis treatment. (5-7) However, the inhibitory activity of isoniazid is reduced by mutations either in InhA or, more commonly, in the KatG catalase-peroxidase responsible for converting the INH prodrug into its active form. (8-10) Thus, identifying inhibitors that directly bind to InhA without the requirement for activation by KatG (direct InhA inhibitors) may represent a valid strategy to overcome isoniazid resistance. (11,12) Hence multiple academic and pharmaceutical efforts have led to the discovery of direct InhA inhibitors. (13-18) However, most of the direct InhA inhibitors so far identified display good InhA inhibitory activity in vitro but poor activity against M. tuberculosis. (13,19-21) Figure 1 Figure 1. NADH-specific reduction of 2-trans-enoyl-ACP catalyzed by InhA (A), in and out conformations of Tyr158 side chain (B). Tyr158 in the in conformation (yellow) in the ternary InhA structure complexed with the C16 substrate analogue THT (trans-2-hexadecenoyl-(N-acetylcysteamine)-thioester (yellow carbon atom) and NAD+ (gray) and Tyr158 in the out conformation (pink) in the binary InhA structure complexed with NAD+ (gray). PDB codes of these structures are 1BVR (3) and 1ENY, (6) respectively. The interactions of InhA with substrate, cofactor, and inhibitors have been extensively studied. (22) One outcome of these investigations is the identification of the active site residue Tyr158 as important both to stabilizing the substrate during the catalytic reaction of M. tuberculosis InhA and to the binding of direct InhA inhibitors. (3,4,23) Two different conformations of the Tyr158 side chain have been identified in binding of direct InhA inhibitors (Figure 1B), an "in" conformation associated with the ternary InhA complex (substrate/cofactor-bound form) and an "out" conformation resembling that observed in the binary InhA complex (cofactor-bound form). (19-24) In the present work, we have applied structure-based virtual screening to select candidate InhA inhibitors from the compound library of the Specs database (www.specs.net), seeking to account for the mobility of Tyr158 by including both conformations of this residue in the screening workflow. This protocol identified two compounds that showed both inhibitory activity against M. tuberculosis cell growth and submicromolar inhibition of purified InhA in in vitro activity assays. A crystal structure for the complex of the most potent of these with InhA identified inhibitor binding in a hydrophobic active site pocket utilized in substrate binding and with Tyr158 in the in conformation. These findings demonstrate that these approaches can identify compounds with InhA inhibitory activity that are active against M. tuberculosis. Further experiments and simulations in progress. |
| Impact | Training in interactive virtual reality, molecular modelling, simulation and structural biology. Exchange visits. Biological assays. Publications: 1: Pakamwong B, Thongdee P, Kamsri B, Phusi N, Taveepanich S, Chayajarus K, Kamsri P, Punkvang A, Hannongbua S, Sangswan J, Suttisintong K, Sureram S, Kittakoop P, Hongmanee P, Santanirand P, Leanpolchareanchai J, Spencer J, Mulholland AJ, Pungpo P. Ligand-Based Virtual Screening for Discovery of Indole Derivatives as Potent DNA Gyrase ATPase Inhibitors Active against Mycobacterium tuberculosis and Hit Validation by Biological Assays. J Chem Inf Model. 2024 Aug 12;64(15):5991-6002. doi: 10.1021/acs.jcim.4c00511. Epub 2024 Jul 12. PMID: 38993154; PMCID: PMC11323271. 2: Kamsri B, Kamsri P, Punkvang A, Chimprasit A, Saparpakorn P, Hannongbua S, Spencer J, Oliveira ASF, Mulholland AJ, Pungpo P. Signal Propagation in the ATPase Domain of Mycobacterium tuberculosis DNA Gyrase from Dynamical- Nonequilibrium Molecular Dynamics Simulations. Biochemistry. 2024 Jun 4;63(11):1493-1504. doi: 10.1021/acs.biochem.4c00161. Epub 2024 May 14. PMID: 38742407; PMCID: PMC11154950. 3: Kamsri B, Pakamwong B, Thongdee P, Phusi N, Kamsri P, Punkvang A, Ketrat S, Saparpakorn P, Hannongbua S, Sangswan J, Suttisintong K, Sureram S, Kittakoop P, Hongmanee P, Santanirand P, Leanpolchareanchai J, Goudar KE, Spencer J, Mulholland AJ, Pungpo P. Bioisosteric Design Identifies Inhibitors of Mycobacterium tuberculosis DNA Gyrase ATPase Activity. J Chem Inf Model. 2023 May 8;63(9):2707-2718. doi: 10.1021/acs.jcim.2c01376. Epub 2023 Apr 19. PMID: 37074047. 4: Thongdee P, Hanwarinroj C, Pakamwong B, Kamsri P, Punkvang A, Leanpolchareanchai J, Ketrat S, Saparpakorn P, Hannongbua S, Ariyachaokun K, Suttisintong K, Sureram S, Kittakoop P, Hongmanee P, Santanirand P, Mukamolova GV, Blood RA, Takebayashi Y, Spencer J, Mulholland AJ, Pungpo P. Virtual Screening Identifies Novel and Potent Inhibitors of Mycobacterium tuberculosis PknB with Antibacterial Activity. J Chem Inf Model. 2022 Dec 26;62(24):6508-6518. doi: 10.1021/acs.jcim.2c00531. Epub 2022 Aug 22. PMID: 35994014. 5: Hanwarinroj C, Thongdee P, Sukchit D, Taveepanich S, Kamsri P, Punkvang A, Ketrat S, Saparpakorn P, Hannongbua S, Suttisintong K, Kittakoop P, Spencer J, Mulholland AJ, Pungpo P. In silico design of novel quinazoline-based compounds as potential Mycobacterium tuberculosis PknB inhibitors through 2D and 3D-QSAR, molecular dynamics simulations combined with pharmacokinetic predictions. J Mol Graph Model. 2022 Sep;115:108231. doi: 10.1016/j.jmgm.2022.108231. Epub 2022 May 28. PMID: 35667143. 6: Hanwarinroj C, Phusi N, Kamsri B, Kamsri P, Punkvang A, Ketrat S, Saparpakorn P, Hannongbua S, Suttisintong K, Kittakoop P, Spencer J, Mulholland AJ, Pungpo P. Discovery of novel and potent InhA inhibitors by an in silico screening and pharmacokinetic prediction. Future Med Chem. 2022 May;14(10):717-729. doi: 10.4155/fmc-2021-0348. Epub 2022 Apr 29. PMID: 35485258. 7: Pakamwong B, Thongdee P, Kamsri B, Phusi N, Kamsri P, Punkvang A, Ketrat S, Saparpakorn P, Hannongbua S, Ariyachaokun K, Suttisintong K, Sureram S, Kittakoop P, Hongmanee P, Santanirand P, Spencer J, Mulholland AJ, Pungpo P. Identification of Potent DNA Gyrase Inhibitors Active against Mycobacterium tuberculosis. J Chem Inf Model. 2022 Apr 11;62(7):1680-1690. doi: 10.1021/acs.jcim.1c01390. Epub 2022 Mar 29. PMID: 35347987. 8: Kamsri P, Hanwarinroj C, Phusi N, Pornprom T, Chayajarus K, Punkvang A, Suttipanta N, Srimanote P, Suttisintong K, Songsiriritthigul C, Saparpakorn P, Hannongbua S, Rattanabunyong S, Seetaha S, Choowongkomon K, Sureram S, Kittakoop P, Hongmanee P, Santanirand P, Chen Z, Zhu W, Blood RA, Takebayashi Y, Hinchliffe P, Mulholland AJ, Spencer J, Pungpo P. Discovery of New and Potent InhA Inhibitors as Antituberculosis Agents: Structure-Based Virtual Screening Validated by Biological Assays and X-ray Crystallography. J Chem Inf Model. 2020 Jan 27;60(1):226-234. doi: 10.1021/acs.jcim.9b00918. Epub 2019 Dec 27. PMID: 31820972. 9: Kamsri P, Punkvang A, Hannongbua S, Suttisintong K, Kittakoop P, Spencer J, Mulholland AJ, Pungpo P. In silico study directed towards identification of the key structural features of GyrB inhibitors targeting MTB DNA gyrase: HQSAR, CoMSIA and molecular dynamics simulations. SAR QSAR Environ Res. 2019 Nov;30(11):775-800. doi: 10.1080/1062936X.2019.1658218. PMID: 31607177. 10: Punkvang A, Kamsri P, Mulholland A, Spencer J, Hannongbua S, Pungpo P. Simulations of Shikimate Dehydrogenase from Mycobacterium tuberculosis in Complex with 3-Dehydroshikimate and NADPH Suggest Strategies for MtbSDH Inhibition. J Chem Inf Model. 2019 Apr 22;59(4):1422-1433. doi: 10.1021/acs.jcim.8b00834. Epub 2019 Mar 14. PMID: 30840825. |
| Start Year | 2015 |
| Description | Catalysis Hub |
| Organisation | Research Complex at Harwell |
| Department | UK Catalysis Hub |
| Country | United Kingdom |
| Sector | Public |
| PI Contribution | Modelling and simulation of enzyme mechanisms for applications in biocatalysts via the Catalysis Hub |
| Collaborator Contribution | Modelling and simulation of enzyme mechanisms for applications in biocatalysts via the Catalysis Hub and training of Hub PDRAs. |
| Impact | Catalysis is a core area of contemporary science posing major fundamental and conceptual challenges, while being at the heart of the chemical industry - an immensely successful and important part of the overall UK economy (generating in excess of £50 billion per annum). UK catalytic science currently has a strong presence, but there is intense competition in both academic and industrial sectors, and a need for UK industrial activity to shift towards new innovative areas posing major challenges for the future. In light of these challenges the UK Catalysis Hub endeavours to become a leading institution, both nationally and internationally, in the field and acts to coordinate, promote and advance the UK catalysis research portfolio. With a strong emphasis on effective use of the world-leading facilities on the RAL campus. Structure The project has four mature themes and a fifth theme starting in 2015 , each with a lead investigator as PI - Catalysis by Design (Catlow); Energy (Hardacre); Environment (Hutchings); Chemical Transformations (Davidson) and the new Biocatalysis and Biotransformations (Nick Turner Manchester) - with the design theme based in the Harwell hub. Each theme is supported by £3 - 3.5M EPSRC funding over 5 years and within each theme there are typically six to eight sub-projects funded initially for 2 years, involving collaborative teams working at a variety of sites throughout the UK. Professor Hutchings acts as director of the whole national programme for the first three year period and chairs the management group, which is supported by a steering group and an industrial advisory panel. We note that engagement with industry is one of the key aims of the catalysis hub project. As well as hosting the design theme, the centre within the Research Complex at Harwell (RCaH) will coordinate the programme, be a base for national and international visitors and provide both training and outreach activities. |
| Start Year | 2015 |
| Title | Confidential |
| Description | Confidential |
| IP Reference | Confidential |
| Protection | Patent / Patent application |
| Year Protection Granted | 2024 |
| Licensed | Commercial In Confidence |
| Impact | Confidential |
| Title | CCP-BioSim software for biomolecular simulation |
| Description | BioSimSpace A new software framework to create an interoperability layer around the many software packages that are already embedded within the biosimulation community. BioSimSpace will enable rapid development of workflows between these software packages that can then be used in conjunction with existing workflow software such as Knime, Pipeline Pilot, ExTASY etc. This project is currently in an early phase of development, more information can be found here. FESetup FESetup is a tool to automate the setup of (relative) alchemical free energy simulations like thermodynamic integration (TI) and free energy perturbation (FEP) as well as post-processing methods like MM-PBSA and LIE. FESetup can also be used for general simulation setup ("equilibration") through an abstract MD engine. The latest releases are available from the project web page. Other Software: ProtoMS - a complete protein Monte Carlo free energy simulation package. Sire - a complete python/C++ molecular simulation framework, particularly focussed around Monte Carlo, QM/MM and free energy methods. PCAZIP - a toolkit for compression and analysis of molecular dynamics trajectories. COCO - a tool to enrich an ensemble of structures, obtained e.g. from NMR. Handy Routines for Ptraj/Cpptraj - additional analysis methods for ptraj and cpptraj. |
| Type Of Technology | Software |
| Year Produced | 2016 |
| Open Source License? | Yes |
| Impact | BioSimSpace A new software framework to create an interoperability layer around the many software packages that are already embedded within the biosimulation community. BioSimSpace will enable rapid development of workflows between these software packages that can then be used in conjunction with existing workflow software such as Knime, Pipeline Pilot, ExTASY etc. This project is currently in an early phase of development, more information can be found here. FESetup FESetup is a tool to automate the setup of (relative) alchemical free energy simulations like thermodynamic integration (TI) and free energy perturbation (FEP) as well as post-processing methods like MM-PBSA and LIE. FESetup can also be used for general simulation setup ("equilibration") through an abstract MD engine. The latest releases are available from the project web page. Other Software: ProtoMS - a complete protein Monte Carlo free energy simulation package. Sire - a complete python/C++ molecular simulation framework, particularly focussed around Monte Carlo, QM/MM and free energy methods. PCAZIP - a toolkit for compression and analysis of molecular dynamics trajectories. COCO - a tool to enrich an ensemble of structures, obtained e.g. from NMR. Handy Routines for Ptraj/Cpptraj - additional analysis methods for ptraj and cpptraj. |
| URL | http://www.ccpbiosim.ac.uk |
| Title | Enlighten |
| Description | Protocols and tools to run (automated) atomistic simulations of protein-ligand (e.g. enzyme-substrate) systems. There is further a plugin to the popular (open source) visualisation program PyMOL that can be used to access and run these protocols. Aimed at: - Experimental biochemists/enzymologists interested in gaining detailed insight into protein-ligand / enzyme-substrate complexes. - Biomolecular researchers that would like to perform simulations in a high(er)-throughput fashion, e.g. for testing and hypothesis generation |
| Type Of Technology | Software |
| Year Produced | 2016 |
| Open Source License? | Yes |
| Impact | Development of the software has allowed engaging with (experimental) enzymologists, e.g. through tutorial workshops on simulation of protein-ligand system for non-experts. Free tutorial workshops were held as follows: 6-6-2016 University of Bristol, Bristol, 40 participants (see also: www.ccpbiosim.ac.uk/training-week-day1 ) 23-2-2017 University of Manchester, Manchester, 25 participants (see also: http://www.ukcatalysishub.co.uk/catalysis-events-publication/atomistic_simulation_workshop) Feedback from the workshops was very positive. |
| URL | http://www.github.com/marcvanderkamp/enlighten |
| Title | FESetup |
| Description | FESetup FESetup is a tool to automate the setup of (relative) alchemical free energy simulations like thermodynamic integration (TI) and free energy perturbation (FEP) as well as post-processing methods like MM-PBSA and LIE. FESetup can also be used for general simulation setup ("equilibration") through an abstract MD engine. The latest releases are available from the project web page. |
| Type Of Technology | Software |
| Year Produced | 2017 |
| Impact | FESetup FESetup is a tool to automate the setup of (relative) alchemical free energy simulations like thermodynamic integration (TI) and free energy perturbation (FEP) as well as post-processing methods like MM-PBSA and LIE. FESetup can also be used for general simulation setup ("equilibration") through an abstract MD engine. The latest releases are available from the project web page. |
| Title | jc14269/CCP-vr-portfolio: First release of DL_POLY simulations in Narupa i-MD. |
| Description | First release. |
| Type Of Technology | Software |
| Year Produced | 2020 |
| Impact | As molecular scientists have made progress in their ability to engineer nanoscale molecular structure, we face new challenges in our ability to engineer molecular dynamics (MD) and flexibility. Dynamics at the molecular scale differs from the familiar mechanics of everyday objects because it involves a complicated, highly correlated, and three-dimensional many-body dynamical choreography which is often nonintuitive even for highly trained researchers. We recently described how interactive molecular dynamics in virtual reality (iMD-VR) can help to meet this challenge, enabling researchers to manipulate real-time MD simulations of flexible structures in 3D. In this article, we outline various efforts to extend immersive technologies to the molecular sciences, and we introduce "Narupa," a flexible, open-source, multiperson iMD-VR software framework which enables groups of researchers to simultaneously cohabit real-time simulation environments to interactively visualize and manipulate the dynamics of molecular structures with atomic-level precision. We outline several application domains where iMD-VR is facilitating research, communication, and creative approaches within the molecular sciences, including training machines to learn potential energy functions, biomolecular conformational sampling, protein-ligand binding, reaction discovery using "on-the-fly" quantum chemistry, and transport dynamics in materials. We touch on iMD-VR's various cognitive and perceptual affordances and outline how these provide research insight for molecular systems. By synergistically combining human spatial reasoning and design insight with computational automation, technologies such as iMD-VR have the potential to improve our ability to understand, engineer, and communicate microscopic dynamical behavior, offering the potential to usher in a new paradigm for engineering molecules and nano-architectures. |
| URL | https://zenodo.org/record/3685789 |
| Description | Atomistic Simulation of Biocatalysts for Non-Experts |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Postgraduate students |
| Results and Impact | On February 23rd, a tutorial workshop was held in Manchester, sponsored by the UK catalysis Hub and CCPBiosim and aimed at non-experts to learn about atomistic simulation of biocatalysts. The day included general introduction, the Enlighten simulation protocols & tools, and a "simulation clinic". 25 participants from around the UK (including from industry), with very positive feedback - all having a better idea about atomistic simulation and how to apply it in their research. |
| Year(s) Of Engagement Activity | 2017 |
| URL | https://www.eventbrite.com/e/atomistic-simulation-of-biocatalysts-for-non-experts-tickets-3150736136... |
| Description | Enlighten: Tools for enzyme-ligand modelling |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Professional Practitioners |
| Results and Impact | A tutorial and feedback session was held on June 6th covering Enlighten, protocols and tools for atomistic simulation of protein-ligand systems that I have developed (www.github.com/marcvanderkamp/enlighten), aimed at non-experts (experimentalists/protein crystallographers). 40 participants from around the UK attended (including from industry), with very positive feedback: all reporting increased understanding of simulation, many likely to use the tools etc. |
| Year(s) Of Engagement Activity | 2016 |
| URL | http://www.ccpbiosim.ac.uk/training-week-day1 |