BlueCryo Image Processing Computing Cluster

Imaging biomolecules at the molecular level to visualize their atomic details is essential to understand how they function in health and disease. Among the techniques capable of imaging biomolecules, electron cryo-microscopy (cryo-EM) has emerged as exceptionally powerful. New direct electron detectors, better microscopes and powerful processing software for the images recorded with this new hardware has led to a genuine revolution in cryo-EM. With these advances it is now possible to determine macromolecular structures at near-atomic resolution by cryo-EM, revealing essential molecular detail. High-resolution structural insight into molecular complexes, and also entire cells and even tissues, provide vital understanding of the molecular mechanisms of life.
Until recently, high-resolution cryo-EM was not possible in the South-West due to the lack of a suitable cryo-microscope. In 2017, we answer this unmet need by opening a South West Regional Facility for High-Resolution Cryo-EM at University of Bristol, equipped with state-of-the-art hardware. This will enable us to collect several terabytes of images or movies per day of successful data collection. To fully exploit the potential of this facility, and to maintain and increase our competitive posture, we must complement the state-of-the-art cryo-microscope with equally powerful high-performance computing (HPC) for image processing. Processing of the terabytes of images is so computation-intense that existing HPC infrastructure can by no means match the computation requirements of the new EM user community.
Therefore, to resolve this bottleneck, we ask here for funding of a computing cluster dedicated to image processing of cryo-EM data sets (BlueCryo). BlueCryo will enable researchers at University of Bristol to determine important structures by single particle cryo-EM and tomography, to interpret these structures using cutting-edge molecular modelling and to analyse the functional dynamics of the biological systems by state-of-the-art simulation methods. The BlueCryo HPC cluster will support the work of many researchers to accelerate their ambitious research, many of them funded by the BBSRC. The new resource will be instrumental for the discovery of new mechanisms of gene expression, to study protein transport into and across membranes, to understand the mechanism of secretion systems which play a crucial role in bacterial infections, and to illuminate a wide range of other important molecular processes responsible for biological function (and malfunction) in our cells. Moreover, the HPC cluster will critically support rational design of entirely novel proteins and peptide assemblies with tailor-made function, in vaccinology, drug delivery and drug discovery. Importantly, the new BlueCryo computing resource, together with the cryo-microscope, will further entice researchers from diverse life-science areas to partake in the cryo-EM revolution and generate new and exciting research synergies all across the South-West and beyond.
Our work has broad implications for multiple areas within the BBSRC remit including synthetic biology, basic bioscience underpinning normal human and animal health and infection. The projects cover areas of direct relevance to BBSRC remit and strategy. The proposal includes researchers with a strong track record of BBSRC funding. Early career researchers and a re-entry fellow are part of our team. Clearly, the computing cluster will not only decisively advance the research here proposed, and generate new programmes, synergies and collaborations. It will also enable us to engage in many other important scientific questions and to train new users and students. Thus, we expect to derive significant additional use in many areas of basic and applied research at University of Bristol, the South-West and beyond.

State-of-the-art image processing for high-resolution electron cryo-microscopy (cryo-EM) is very computation-intense. High-performance computing clusters are required to efficiently analyse cryo-EM data sets and solve structures at near-atomic resolution in a reasonable time as well as for their interpretation by molecular modelling, e.g. using Rosetta, and molecular dynamics simulations.
In 2017, the new South-West Regional Facility for High-Resolution Cryo-EM will become operational at the University of Bristol (UoB), outfitted with a FEI Talos Arctica cryo-microscope, energy filter and a Gatan K2 Summit direct electron detector. It will also serve as a feeder instrument into the BBSRC-funded National Facility for Electron Cryo-Microscopy at Harwell with currently two Titan Krios cryo-microscopes. Thanks to automated data collection, terabytes of images and movies can be collected per day. These need to be analysed with efficient throughput to match the speed of data acquisition.
Current HPC resources at UoB cannot provide the required computation time, which is more than 100,000 cpu hours per project, alongside additional time for modelling and simulation for detailed structural interpretation. Therefore, we aim to implement a new cluster, BlueCryo, dedicated to image processing and cryo-EM structure interpretation. The cluster will have a GPU and a CPU component, with sufficient memory and disk space and will provide the equivalent of ~200,000 cpu hours per week. This will allow researchers at UoB (including many BBSRC-funded) to efficiently pursue their research on protein translocation across and into membranes, bacterial secretion systems, transcription factors, heart muscle filaments and target of rapamycin complexes as well as designed synthetic peptide assemblies for drug delivery vehicles and vaccines development. The interest to apply cryo-EM and image processing is paramount with many research groups keen to benefit from the cryo-EM revolution.

The first and foremost impact of this proposal is the successful establishment of electron cryo-microscopy image processing at the University of Bristol (UoB). The proposed high-performance computing (HPC) cluster dedicated to image processing will decisively accelerate the ambitious research programmes pursued by leading researchers at the University. Moreover, the resource will enable many other laboratories to partake in the cryo-EM revolution, to advance their research and discovery.
The resource will furthermore enable us to develop new methodology to address key questions in the field leading to ample academic benefit beyond the ambitious research questions we address. Of note, the HPC cluster will be particularly instrumental to advance and consolidate the careers of early career and re-integrating researchers. Moreover, the proposed BlueCryo cluster will positively impact on the career development of the System Administrator employed to manage it, potentiated by the excellent support from Advanced Computing Research Centre (ACRC).
The individual research projects within the team possess significant potential for impact in the longer term. Our work addresses crucial questions which are at the core of biological function in prokaryotes and eukaryotes including humans. Our research programmes underpin our understanding of the healthy organism, disease-related changes and infection strategies used by pathogens. Work within the proposal seeks to generate new cellular effectors, versatile vehicles for drug delivery and next-generation vaccines by designing novel proteins and synthetic peptide assemblies. Through informing our basic understanding of essential cellular processes, it is most likely that our work will inform long-term projects in other fields reaching through to the discovery of new and better drugs in the pharma and biotech industries. Potential translational applications of our work will be identified from within the labs involved as well as by continuing liaison with our Research and Enterprise Department. Any outcomes of this work that are exploitable, notably in terms of intellectual property or knowledge transfer to the private sector, are handled by the highly experienced team within RED; who engage closely with funders such as BBSRC when appropriate.
The projects include considerable opportunity to train the researchers involved in areas that go beyond the day-to-day research methodology. Examples include our extensive integration with public communication and outreach programmes, the extensive network of University schemes to benefit the training and development of research staff (Bristol is at the forefront of research staff development). The environment as a whole at UoB is highly conducive to career development of our staff beyond academic, basic science research alone and thus contributes to the economic development of the nation.
The projects are all very data intensive and the management and analysis of such large (terabyte) datasets is applicable to many areas of professional life, outside of life science. In collaboration with the Bristol ACRC, the projects offer the researchers to develop and strengthen their capabilities in high-performance computing, script writing and bioinformatics; competencies which are high in demand on the job market.
This work will lead to significant image data that is readily used in both public understanding of science and artistic arenas. Through our public engagement plans and other outreach activities, this work therefore is likely to contribute to local exhibitions, promotions or displays as has been the case with previous work from the applicants' labs and others within UoB.