Transforming synthetic drug manufacturing: novel processes, methods and tools
Lead Research Organisation:
Imperial College London
Department Name: Chemical Engineering
Abstract
The pharmaceutical industry, a key player in UK manufacturing, faces huge challenges in turning promising new molecules into affordable medicines. While synthetic drugs make up the largest part of pharmaceutical companies' drug portfolios, with peptides representing an increasingly important class of drugs, the road from the discovery of a drug molecule to a commercial product that benefits patients remains frustratingly long and arduous, with the total cost of development reaching $2.6bn and 10 years per new chemical entity (NCE) . Manufacturing and formulation can be a rate-limiting and costly step due to the difficulty in achieving the required molecular precision or yield in synthesis, in ensuring high yields during purification, and in producing final products with the stability and efficacy that maximise patient benefit. Furthermore, the pharmaceutical industry is facing a drive to improve the economic and environmental performance of its manufacturing processes, which currently suffer from extremely low material efficiency, with factors of 0.01 to 0.1 not unusual , the production of a large amount of waste and a slow adoption of quality by design (QbD) concepts, especially for more complex products.
This Prosperity Partnership builds on an existing collaboration between leading industrial and academic investigators to address critical issues in our ability to manufacture synthetic drugs in a cost and time effective way. Together, we have identified scientific hurdles that prevent the successful manufacture and delivery to patients of key medicines and we have devised an ambitious research programme to overcome them. A unique and exciting feature of our approach is to draw on expertise and advances in the manufacture of small molecules to enable radical progress in the synthetic manufacture of much larger peptide drugs, considering the entire chain from drug substance to drug product. Our programme will thus deliver fundamental understanding, models, technologies and design methodologies in order to accelerate the synthesis, isolation, purification and formulation of synthetic drugs of varying sizes, from small molecules to peptides, and to push the boundary of feasibility in relation to peptide drugs. Beyond its scientific achievements, the Prosperity Partnership will positioning the UK at the leading edge of expertise and innovation in the manufacturing of high-value synthetic drugs, contributing to the growth of a value-creating innovation ecosystem.
Eli Lilly and the two academic partners have co-created a comprehensive research programme with the ambition to reduce radically the cost, time and risk inherent in the manufacturing of synthetic drugs, bringing health and economic benefits to the UK.
Our research vision is thus to deliver novel systems-based engineering design methods for the rapid development of manufacturing processes for advanced synthetic drugs and drug products, strongly rooted in scientific understanding and building on state-of-the-art manufacturing technologies, explainable AI ,modelling and experimental approaches.
Our programme has been designed around 5 interacting work packages
1. Novel synthesis methods for drug substances (active ingredients), including complex peptides which are a very promising emerging therapy
2. Advanced techniques for drug substance crystallisation based on fundamental thermodynamic modelling
3. Advanced techniques for drug substance purification, including the emerging area of peptide chromatography
4. Advanced manufacturing and stability analysis of drug products. Drug substances must be formulated as drug products which must be proven to be stable over their shelf life. Here will explore the interactions between design, manufacturing and stability.
5. Cross-cutting systems engineering methods for model-based design and operational optimisation
This Prosperity Partnership builds on an existing collaboration between leading industrial and academic investigators to address critical issues in our ability to manufacture synthetic drugs in a cost and time effective way. Together, we have identified scientific hurdles that prevent the successful manufacture and delivery to patients of key medicines and we have devised an ambitious research programme to overcome them. A unique and exciting feature of our approach is to draw on expertise and advances in the manufacture of small molecules to enable radical progress in the synthetic manufacture of much larger peptide drugs, considering the entire chain from drug substance to drug product. Our programme will thus deliver fundamental understanding, models, technologies and design methodologies in order to accelerate the synthesis, isolation, purification and formulation of synthetic drugs of varying sizes, from small molecules to peptides, and to push the boundary of feasibility in relation to peptide drugs. Beyond its scientific achievements, the Prosperity Partnership will positioning the UK at the leading edge of expertise and innovation in the manufacturing of high-value synthetic drugs, contributing to the growth of a value-creating innovation ecosystem.
Eli Lilly and the two academic partners have co-created a comprehensive research programme with the ambition to reduce radically the cost, time and risk inherent in the manufacturing of synthetic drugs, bringing health and economic benefits to the UK.
Our research vision is thus to deliver novel systems-based engineering design methods for the rapid development of manufacturing processes for advanced synthetic drugs and drug products, strongly rooted in scientific understanding and building on state-of-the-art manufacturing technologies, explainable AI ,modelling and experimental approaches.
Our programme has been designed around 5 interacting work packages
1. Novel synthesis methods for drug substances (active ingredients), including complex peptides which are a very promising emerging therapy
2. Advanced techniques for drug substance crystallisation based on fundamental thermodynamic modelling
3. Advanced techniques for drug substance purification, including the emerging area of peptide chromatography
4. Advanced manufacturing and stability analysis of drug products. Drug substances must be formulated as drug products which must be proven to be stable over their shelf life. Here will explore the interactions between design, manufacturing and stability.
5. Cross-cutting systems engineering methods for model-based design and operational optimisation
Planned Impact
The outputs of this prosperity partnership between Eil Lilly, Imperial, and UCL will have the potential to impact the pharmaceutical industry directly, as well as other industries and society. For example, the WP1 outputs, which will reduce cost and time-to market of drug substance synthesis, can produce not only economic but also environmental and broader societal benefits.
It will deliver novel methodologies for (i) drug substance synthesis (ii) drug substance purification and isolation (iii) drug product formulation and stability, and (iv) underpinning process systems engineering methods embodied in a range of tools. These will be demonstrated on a range of industrial case studies and demonstration projects supported by our partner.
The invention and development of such methods and their adoption by industry has the potential to lead to a step change in the manufacturing and quality control of advanced therapeutics. The new synthesis techniques for example will for the first time open up opportunities for larger peptide based therapies. The existing underpinning systems and modelling techniques have been shown in to have a 10-fold return in investment in the sector (Am Ende et al., AIChE An, 2010); We expect the next generation of tools to have at least the same impact.
The beneficiaries include:
Pharma industry and related academia: The recent UK Life Sciences Sector Deal states "The life sciences industry is one of the most important pillars of the UK economy, contributing over £70 billion a year and 240,000 jobs across the country" and highlights "advanced therapies" as one of the key platforms. We have aligned our programme with exactly this opportunity. The integrated work packages have the overarching objective of bringing cost-effective new therapies to market as quickly as possible.
Academics and software companies working on model-based systems engineering: The beneficiaries will be those working in the area of design of experiments, quality by design, advanced thermodynamics, optimization and process analytics and control.
Wider industrial partners: The methods are generic and can be applied to a broad range of problems. The tools will have impact in the wider manufacturing sector ranging from consumer goods to agrochemicals and oil and gas; CPSE's industrial consortium includes representatives from these sectors.
Society: The improvements that will arise from our advancements should support faster development and manufacturing of advanced therapies. Our platform will reduce the time and cost of development for new drug production processes.
We shall ensure effective impact outcomes through:
Training: The researchers will gain career-relevant skills and will benefit from the cross-disciplinary interactions within the team as well as with the industrial partner, including secondment opportunities. Senior representatives from Eli Lilly will act as co-supervisors and provide advice, materials, data and analytical resources. They will also provide input on the quality, timeliness and relevance of the methods developed.
Software dissemination: A range of software tools will be developed in the course of the research and made available to the academic community to download and apply to their models.
Manufacturing technologies: we anticipate new manufacturing technologies (e.g. Nanostar sieving for liquid phase synthesis).
Project website and workshops: The core outputs developed herein will be advertised through a website dedicated to the project. We shall also hold dissemination workshops for end users and regulator representatives near the project end
It will deliver novel methodologies for (i) drug substance synthesis (ii) drug substance purification and isolation (iii) drug product formulation and stability, and (iv) underpinning process systems engineering methods embodied in a range of tools. These will be demonstrated on a range of industrial case studies and demonstration projects supported by our partner.
The invention and development of such methods and their adoption by industry has the potential to lead to a step change in the manufacturing and quality control of advanced therapeutics. The new synthesis techniques for example will for the first time open up opportunities for larger peptide based therapies. The existing underpinning systems and modelling techniques have been shown in to have a 10-fold return in investment in the sector (Am Ende et al., AIChE An, 2010); We expect the next generation of tools to have at least the same impact.
The beneficiaries include:
Pharma industry and related academia: The recent UK Life Sciences Sector Deal states "The life sciences industry is one of the most important pillars of the UK economy, contributing over £70 billion a year and 240,000 jobs across the country" and highlights "advanced therapies" as one of the key platforms. We have aligned our programme with exactly this opportunity. The integrated work packages have the overarching objective of bringing cost-effective new therapies to market as quickly as possible.
Academics and software companies working on model-based systems engineering: The beneficiaries will be those working in the area of design of experiments, quality by design, advanced thermodynamics, optimization and process analytics and control.
Wider industrial partners: The methods are generic and can be applied to a broad range of problems. The tools will have impact in the wider manufacturing sector ranging from consumer goods to agrochemicals and oil and gas; CPSE's industrial consortium includes representatives from these sectors.
Society: The improvements that will arise from our advancements should support faster development and manufacturing of advanced therapies. Our platform will reduce the time and cost of development for new drug production processes.
We shall ensure effective impact outcomes through:
Training: The researchers will gain career-relevant skills and will benefit from the cross-disciplinary interactions within the team as well as with the industrial partner, including secondment opportunities. Senior representatives from Eli Lilly will act as co-supervisors and provide advice, materials, data and analytical resources. They will also provide input on the quality, timeliness and relevance of the methods developed.
Software dissemination: A range of software tools will be developed in the course of the research and made available to the academic community to download and apply to their models.
Manufacturing technologies: we anticipate new manufacturing technologies (e.g. Nanostar sieving for liquid phase synthesis).
Project website and workshops: The core outputs developed herein will be advertised through a website dedicated to the project. We shall also hold dissemination workshops for end users and regulator representatives near the project end
Publications
Adjiman C
(2021)
Process Systems Engineering Perspective on the Design of Materials and Molecules
in Industrial & Engineering Chemistry Research
Al Musaimi O
(2022)
Strategies for Improving Peptide Stability and Delivery.
in Pharmaceuticals (Basel, Switzerland)
Al Musaimi O
(2022)
Greener Cleavage of Protected Peptide Fragments from Sieber Amide Resin.
in ChemistryOpen
Al Musaimi O
(2023)
Successful synthesis of a glial-specific blood-brain barrier shuttle peptide following a fragment condensation approach on a solid-phase resin.
in Journal of peptide science : an official publication of the European Peptide Society
Al Musaimi O
(2023)
Prediction of peptides retention behavior in reversed-phase liquid chromatography based on their hydrophobicity.
in Journal of separation science
Al Musaimi O
(2023)
Factors Influencing the Prediction Accuracy of Model Peptides in Reversed-Phase Liquid Chromatography
in Separations
Bascone D
(2020)
Hybrid Mechanistic-Empirical Approach to the Modeling of Twin Screw Feeders for Continuous Tablet Manufacturing
in Industrial & Engineering Chemistry Research
Beran GJO
(2022)
How many more polymorphs of ROY remain undiscovered.
in Chemical science
Bernet T
(2024)
Modeling the Thermodynamic Properties of Saturated Lactones in Nonideal Mixtures with the SAFT-? Mie Approach.
in Journal of chemical and engineering data
Besenhard M
(2021)
Recent advances in modelling and control of liquid chromatography
in Current Opinion in Chemical Engineering
Description | The key findings impact every aspect of pharmaceutical process development: In drug substance synthesis, we have developed novel strategies to controllably manipulate membrane performance (permeability and selectivity) by various means including dope composition, drying conditions and surface grafting; this can increase yield/reduce waste and cost in manufacturing. We have achieved mechanistic understanding of hydrogen cyanide formation in amino acid activation and a systematic in silico method to identify reaction solvents that suppress this unsafe byproduct, to increase the safety of peptide synthesis. We have developed computer-aided design approaches to identify synthesis routes with better yield and reduced environmental impact, through solvent design or build design (linear vs fragment-based). We have also advanced the design and synthesis of novel hubs for peptide synthesis, and developed computational methods to predict the solubility of hubs, accelerating the design process. In drug substance purification isolation via crystallisation, good progress has been made on experimental and computational methods to accelerate crystallisation design. We have developed new experimental protocols to measure accurately the solubility of short chain peptides; generated new data for dipeptides and homopeptides with insights on effects of chain length, solvent, sequence; identified new crystals of GG353 and some of its fragments identified for the first time, although such crystals had been elusive in previous studies; produced the first successful predictions of the solubility of several amino acids and dipeptides in water and alcohols, validated against experimental data from the literature and from the project. We have in addition developed a modelling framework for charged active pharmaceutical ingredients, proposing new theory and validating the application to pH effects on the solubility of pharmaceuticals. We have uncovered a new phase stability criterion for charged systems and used this to propose new algorithms for phase stability analysis in charged and reacting mixtures and parameter estimation for such systems. This has applications much beyond pharmaceuticals. Through algorithmic development, we have achieved reductions of up to 60% in computational time in crystal structure prediction (CSP) by exploiting the mathematical and physical properties of crystals. Finally, we have proposed novel method for screening co-crystallising agents developed and applied them successfully to three pharmaceutical compounds in a blind context. In drug substance isolation and purificatio via high pressure liquid chromatography (HPLC), we have combined experimental and computational methods to propose a new workflow for HPLC design, that can improve product quality and reduce the time and cost of design. This includes creating a digital twin of HPLC for molecules/analytes, using mechanistic and ML models and data to predict retention. This has been validated, in excellent agreement with operational data. We have proposed a superstructure optimisation strategy built to optimise preparative scale operation implement and tested it on initial case studies. We have put forward a strategy to extend work on small moelcules to RP-HPLC for larger molecules (peptides) by integrating peptide molecular properties. Finaly, we have proposed a protocol for advanced experimental characterisation of RPLC stationary phases and column hydrophobicity. In drug product manufacture, stability and delivery, we have developed new mathematical models of powder flow and mixing in continuous screw mixers, implementing rheology models and mixing/segregation models, in excellent agreement with data. We have generated new data from experimental studies of peptide aggregation behaviour using dynamic light scattering and size exclusion chromatography for model peptides. We have successfully tested the effectiveness of new functionalised media for the immobilisation of lipid layers and enzymes onto chromatographic media. Finally, we have developed a set of underpinning systems technologies to support all of the above activities. This includes an improved automated decision-making processes for the generation and maintenance of the chemometric models used in process analytical technology; an efficient approach to identify probabilistic design spaces for complex and highly nonlinear models; a new approach for the design of experiments subject to conflicting objectives, which has so far been tested on simple examples. |
Exploitation Route | The outcomes are in the form of methods, software and data, that can all be used by others to accelerate process development. All approaches are applicable to small molecules or peptides, but can also be extended to other new modalities such as oligonucleotides, a fast growing type of therapy. In addition, many of the approaches can be used in other contexts, e.g., battery modelling, CO2 capture process design etc. |
Sectors | Chemicals Energy Manufacturing including Industrial Biotechology Pharmaceuticals and Medical Biotechnology |
Description | The timely delivery of novel medicines to patients requires the rapid development of new manufacturing processes that offer reliability in terms of product safety, quality and uninterrupted supply. This is challenging due to the lack of experimental data during process development and the time pressures arising from the need for material for clinical trials. In recent years, the fast growth of new modalities such as peptides and oligonucleotides has provided new hope for patients while creating additional challenges for process development. The overarching aim of this partnership is to develop and demonstrate a suite of new methods that can support effective and safe process development, from synthesis to separation, through the design of minimal, yet highly informative, experimental campaigns. More specifically, a key objective is to develop new protocols and systems engineering tools to design experimental campaigns that provide the data necessary for process development, while reducing the number of experiments required and hence costs. The experiments designed should inherently support a range of objectives in medicines manufacturing such as reduced environmental impacts, increased safety and increased manufacturing resilience. Through our industry-academia partnership, we have made significant advances in the design of smarter experiments through a multi-faceted programme of research that involves: • the consideration of multiscale, molecule-to-process, interactions, • the adoption of systems thinking to develop approaches that are transferable to many aspects of process development and to other sectors, • the use of multidisciplinarity perspectives to develop industrially-relevant approaches that build on a fundamental understanding of the underpinning science. Our broad-ranging partnership has involved deep and sustained collaboration between industry and academic partners. This has resulted in significant benefits in the design of experimental campaigns, where we have made four key contributions: (i) A novel approach to design experiments that optimally explore the process space, facilitating scale-up at minimal cost. The experiments have much greater likelihood of feasibility and information content than a standard (factorial) experiment design. This approach has been deployed successfully within Lilly. It is now commonly used across drug product projects, ensuring a reliable supply of clinical material by essentially eliminating operational failures at commercial scale. For instance, for a batch wet granulation process, the approach was instrumental in achieving scale-up despite limited material that meant only 3 experiments could be conducted at lab scale. These contained sufficient information to parameterise the model reliably and to increase process scale 60-fold. This approach has also been applied in a "live" process development project for a new continuous RNA manufacturing process. It was very useful in identifying counter-intuitive experiment designs which a standard factorial approach would miss due to complex, non-monotonic behaviour. (ii) A new algorithmic framework for computing safe and robust experimental campaigns in the presence of modelling uncertainties. One key innovation is a bi-objective optimisation formulation accounting for uncertainty when designing a maximally-informative campaign, while reducing the risk of designing non-informative experiments by considering worst-case scenarios. Another capability is to design an optimal campaign that satisfies critical safety/operational constraints with a given probability. This framework is available through the open-source code Pydex, leveraging state-of-the-art convex optimisation technology. An implementation is also underway in the Siemens gPROMS modelling environment, a commercial tool widely used across the pharmaceutical and numerous other sectors. (iii) A computer-based methodology to identify promising solvents for synthesis and crystallisation for safe targeted experiments, which takes into account multiple criteria to enable the integration of sustainability and safety in early-stage process development (cf Q4 and Q5). (iv) A workflow to support crystallisation development for peptide manufacturing. It is notoriously difficult to determine whether a peptide can crystallise and under what conditions. Our novel screening approaches, based on understanding of molecular interactions, has led to previously-elusive crystallisation of 12 fragments (out of 17) of a 39-mer Lilly peptide, GG353, and two novel crystalline forms. The knowledge gained from studying the liquid-liquid phase separation of peptides and the introduction of a micro-seeding approach have been instrumental in developing a robust crystallisation route for GG353. Understanding the crystallisability of peptides improves manufacturability, enabling higher throughput, lowering product costs, increasing product quality and stability, including opportunities for novel formulations and technologies. Beyond the industrial relevance of these newly-discovered crystal forms, these data will be incorporated into a model to understand how sequence impacts crystallisability. This is critically important in process development as the knowledge of which fragments can crystallise impacts decisions on process configuration and synthetic route. This work has resulted in a new workflow for crystallisation which is being deployed at Lilly for new candidate molecules. Our systems approach supports our focus on sustainability. In model-based design, we incorporate constraints and a range of key performance indicators as objectives within the methodologies developed, including cost, environment and safety. We have held numerous discussions to identify appropriate metrics for process development, such as safety and environmental indicators, E-factors, energy consumption. Because solvents are a major source of environmental and safety impacts for the pharmaceutical industry, we have developed methods to support early decision making (e.g., route selection, solvent selection for individual process units) based on a holistic end-to-end analysis of the process options. The benefits of this approach, in terms of solvent or process E-factor and energy consumption, have been highlighted on case studies on mefenamic acid, ibuprofen and peptide production. Several objectives and key performance indicators used in our work have further impacts on the social dimension of pharmaceutical process development. These include efforts to accelerate process development and thus reduce the time needed for new therapies to reach patients and minimising the cost of medicines through reduced R&D and manufacturing costs. More broadly, we have been training a new generation highly-skilled engineers who can help to ensure the sustainability of the pharmaceutical sector in the UK. To date eighteen researchers have moved into new roles following their time within the partnership. Nine of these have moved within academia, including several to independent academic positions, a further eight have moved into roles within the pharmaceuticals and associated sectors, and one has moved into financial services. |
First Year Of Impact | 2022 |
Sector | Chemicals,Energy,Manufacturing, including Industrial Biotechology,Pharmaceuticals and Medical Biotechnology |
Impact Types | Economic |
Description | ADOPT - Advancing optimisation technologies through international collaboration |
Amount | £1,344,649 (GBP) |
Funding ID | EP/W003317/1 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 02/2022 |
End | 02/2026 |
Description | ATLAS - Automated high-throughput platform suite for accelerated molecular systems discovery |
Amount | £1,281,110 (GBP) |
Funding ID | EP/V029142/1 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 05/2021 |
End | 06/2024 |
Description | BASF / RAEng Research Chair in Data-Driven Optimisation |
Amount | £216,000 (GBP) |
Organisation | Royal Academy of Engineering |
Sector | Charity/Non Profit |
Country | United Kingdom |
Start | 03/2022 |
End | 02/2027 |
Description | BioSMART: BIOreactor Spatial Mapping and Actuation in Real Time |
Amount | £1,011,858 (GBP) |
Funding ID | EP/W024969/1 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 02/2023 |
End | 08/2025 |
Description | CASE Studentship - D Pessina |
Amount | £31,500 (GBP) |
Organisation | AstraZeneca |
Sector | Private |
Country | United Kingdom |
Start | 09/2022 |
End | 09/2026 |
Description | DEVELOPMENT OF A MACHINE LEARNING-ASSISTED DIGITAL TWIN PLATFORM FOR REAL-TIME OPTIMISATION OF REACTION SYSTEMS UNDER UNCERTAINTY |
Amount | £714,354 (GBP) |
Funding ID | EP/X024016/1 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 08/2023 |
End | 09/2026 |
Description | Development and Application of Probabilistic Quantification within Pharmaceutical Manufacturing |
Amount | £81,400 (GBP) |
Funding ID | 2767857 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 08/2020 |
End | 03/2024 |
Description | Digitalisation and automation of high-value biomanufacturing |
Amount | £754,778 (GBP) |
Funding ID | EP/X024156/1 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 03/2023 |
End | 03/2026 |
Description | EU MSCA International Fellowship |
Amount | € 225,000 (EUR) |
Funding ID | 101026339 |
Organisation | Marie Sklodowska-Curie Actions |
Sector | Charity/Non Profit |
Country | Global |
Start | 09/2021 |
End | 09/2023 |
Description | Nanostar Sieving for Oligonucleotides Manufacture (NanoSieveOligo) |
Amount | £486,259 (GBP) |
Funding ID | EP/T00827X/2 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 05/2021 |
End | 06/2023 |
Description | Precise deposition of complex particles for structured functional products |
Amount | £218,888 (GBP) |
Funding ID | EP/V003070/2 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 02/2022 |
End | 09/2024 |
Description | Predicting the thermodynamic stability of multicomponent solids |
Amount | £40,500 (GBP) |
Organisation | Syngenta International AG |
Department | Syngenta Ltd (Bracknell) |
Sector | Private |
Country | United Kingdom |
Start | 09/2021 |
End | 03/2025 |
Description | Prosperity Partnership Call 3 Strategic Students-Eli Lilly and Imperial College London |
Amount | £657,409 (GBP) |
Funding ID | EP/T518207/1 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 09/2020 |
End | 10/2025 |
Description | SCALING-UP OF A HIGHLY MODULAR ROTATING PACKED BED PLANT WITH AN EFFICIENT SOLVENT FOR CAPTURE COST REDUCTION |
Amount | € 6,330,101 (EUR) |
Funding ID | 101075727 |
Organisation | European Commission H2020 |
Sector | Public |
Country | Belgium |
Start | 11/2022 |
End | 10/2026 |
Description | SynHiSel |
Amount | £7,328,275 (GBP) |
Funding ID | EP/V047078/1 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 02/2022 |
End | 01/2027 |
Description | System Builders - Device Assembly from Nanoporous Materials Developed from Current Platform Grant (EP/J014974/1) |
Amount | £524,052 (GBP) |
Funding ID | EP/R029180/2 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 06/2021 |
End | 06/2023 |
Title | A predictive group-contribution framework for the thermodynamic modelling of CO2 absorption in cyclic amines, alkyl polyamines, alkanolamines and phase-change amines: new data and SAFT- gamma Mie parameters.FPE 2022 |
Description | All data in the figures in the publication. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | Enabled the prediction of solubility of pharmaceutical compounds |
URL | https://zenodo.org/record/7291413 |
Title | AcRoPLS - Accurate Robust PLS (Chryssoula Kappatou) |
Description | AcRoPLS (Acurate Robust PLS) provides a methodology to create accurate and robust PLS models based on solving a data-driven optimization problem that couples data pre-processing and model regression to a single optimization step. The accuracy objective is evaluated based on the performance of the generated model on predicting the model output on a test set. For the robustness objective, a novel metric based on the method of moments applied for different realizations of a known variability source evaluated again on a test set. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | The dataset provides a methodology to create accurate and robust PLS models based on solving a data-driven optimization problem that couples data pre-processing and model regression to a single optimization step. |
URL | https://github.com/ckappatou/AcRoPLS |
Title | An approach for modelling simultaneous fluid-phase and chemical reaction equilibria in multicomponent systems via Lagrangian duality: The reactive HELD algorithm. |
Description | This is a data set associated with the paper An approach for modeling simultaneous fluid-phase and chemical reaction equilibria in multicomponent systems via Lagrangian duality: The reactive HELD algorithm. by Felipe A. Perdomo, George Jackson, Amparo Galindo, Claire S. Adjiman. The manuscript is presented as a proceeding of the 33rd European Symposium on Computer-Aided Process Engineering (ESCAPE33), June 18-21, 2023, in Athens, Greece. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | Conference presentation; used for project in pharmaceutical development and CO2 capture |
URL | https://zenodo.org/record/7418447 |
Title | Bootstrap PLS (James Odgers) |
Description | Files contain methods to produce Design Space (DS) and probabilistic predictions. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | No |
Impact | The file contains all of the methods used to make probabilistic predictions. |
URL | https://github.com/jamesacodgers/bootstrapPLS |
Title | Computer-aided solvent mixture design for the crystallisation and isolation of mefenamic acid |
Description | The files contain the MINLP formulations presented in this publication. The MINLP problems are implemented and solved in GAMS version 28.2.0, using SBB, a local branch-&-bound MINLP solver. |
Type Of Material | Database/Collection of data |
Year Produced | 2020 |
Provided To Others? | Yes |
Impact | Application to pharmaceutical process development |
URL | https://zenodo.org/record/3628747 |
Title | Dataset for "Integrating model-based design of experiments and computer-aided solvent design" |
Description | Dataset accompanying the paper "Integrating model-based design of experiments and computer-aided solvent design". See https://www.sciencedirect.com/science/article/pii/S0098135423002156. |
Type Of Material | Database/Collection of data |
Year Produced | 2023 |
Provided To Others? | Yes |
Impact | None yet |
URL | https://zenodo.org/doi/10.5281/zenodo.7839710 |
Title | Dataset for "Solvent design assisted by mechanistic insights: methods and application to peptide synthesis" |
Description | Dataset accompanying the PhD thesis "Solvent design assisted by mechanistic insights: methods and application to peptide synthesis" |
Type Of Material | Database/Collection of data |
Year Produced | 2024 |
Provided To Others? | Yes |
Impact | None yet |
URL | https://zenodo.org/doi/10.5281/zenodo.8396100 |
Title | Description of the thermodynamic properties and fluid-phase behaviour of aqueous solutions of linear, branched, and cyclic amines. AIChE J 2021 |
Description | All computational data for figures presented in the publication. |
Type Of Material | Database/Collection of data |
Year Produced | 2021 |
Provided To Others? | Yes |
Impact | Use in the modelling of complex mixtures of fluids, including in pharmaceutical process development and in CO2 capture. |
URL | https://zenodo.org/record/4488482 |
Title | ENTMOOT: A Framework for Optimization over Ensemble Tree Models |
Description | Gradient boosted trees and other regression tree models perform well in a wide range of real-world, industrial applications. These tree models (i) offer insight into important prediction features, (ii) effectively manage sparse data, and (iii) have excellent prediction capabilities. Despite their advantages, they are generally unpopular for decision-making tasks and black-box optimization, which is due to their difficult-to optimize structure and the lack of a reliable uncertainty measure. ENTMOOT is our new framework for integrating (already trained) tree models into larger optimization problems. The contributions of ENTMOOT include: (i) explicitly introducing a reliable uncertainty measure that is compatible with tree models, (ii) solving the larger optimization problems that incorporate these uncertainty aware tree models, (iii) proving that the solutions are globally optimal, i.e. no better solution exists. In particular, we show how the ENTMOOT approach allows a simple integration of tree models into decision-making and black-box optimization, where it proves as a strong competitor to commonly-used frameworks. |
Type Of Material | Computer model/algorithm |
Year Produced | 2022 |
Provided To Others? | No |
Impact | The contributions of ENTMOOT include: (i) explicitly introducing a reliable uncertainty measure that is compatible with tree models, (ii) solving the larger optimization problems that incorporate these uncertainty aware tree models, (iii) proving that the solutions are globally optimal, i.e. no better solution exists. In particular, we show how the ENTMOOT approach allows a simple integration of tree models into decision-making and black-box optimization, where it proves as a strong competitor to commonly-used frameworks. |
URL | https://www.researchgate.net/publication/351340218_ENTMOOT_A_framework_for_optimization_over_ensembl... |
Title | Local minimization results of 107 small organic hydrates |
Description | We present a systematic evaluation of a state-of-the-art crystal structure prediction (CSP) force field method for organic hydrates. A total number of 107 hydrates are minimized locally with six models with different functionals, basis sets and/or the use of a continuum polarizability correction). The geometric differences between experimental structures and the corresponding minimization outputs are compared in terms of atomic positions as measured by the root-mean-squared deviation. The CPU time required is also investigated. Five of the six models are found to give a good degree of accuracy in more than 95% of cases, but with varying computational costs. |
Type Of Material | Database/Collection of data |
Year Produced | 2023 |
Provided To Others? | Yes |
Impact | None yet |
URL | https://zenodo.org/record/7978986 |
Title | Model-Based Solvent Selection for the Synthesis and Crystallisation of Pharmaceutical Compounds |
Description | The pharmaceutical industry needs design tools to identify greener and more resource-efficient process routes. Current model-based solvent selection methodologies often focus on the choice of solvent in a single unit operation, with fixed operating conditions. In particular, the two key stages of synthesis and separation are usually treated independently. This often results in the use of different solvents for each processing task, which typically requires energy-intensive solvent swap operations. In the current paper, we present a novel computer-aided approach based on computer-aided mixture/blend design (CAMbD) that couples property prediction with simple process models and optimisation to simultaneously identify optimal solvents and anti-solvents, compositions and process conditions for integrated synthesis and crystallisation. Solvents are chosen using key performance indicators (KPIs) that quantify mass efficiency and product quality. The proposed methodology is illustrated by identifying promising reaction and crystallisation solvents for the synthesis of mefenamic acid from 2,3-dimethylaniline and 2-chlorobenzoic acid. Furthermore, multi-objective optimisation is deployed to highlight the trade-offs between the solvent or process E-factor and safety indicators, and between the solvent E-factor and crystal yield. The inclusion of mass-based KPIs and safety specifications ensures that only high-performance solvents are chosen. The findings of our approach are expected to guide the rational selection of solvents for greener pharmaceutical manufacturing during early-stage process development. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | Paper; Use in pharmaceutical process development projects |
URL | https://zenodo.org/record/5555796 |
Title | Model-Based Solvent Selection for the Synthesis and Crystallisation of Pharmaceutical Compounds |
Description | The pharmaceutical industry needs design tools to identify greener and more resource-efficient process routes. Current model-based solvent selection methodologies often focus on the choice of solvent in a single unit operation, with fixed operating conditions. In particular, the two key stages of synthesis and separation are usually treated independently. This often results in the use of different solvents for each processing task, which typically requires energy-intensive solvent swap operations. In the current paper, we present a novel computer-aided approach based on computer-aided mixture/blend design (CAMbD) that couples property prediction with simple process models and optimisation to simultaneously identify optimal solvents and anti-solvents, compositions and process conditions for integrated synthesis and crystallisation. Solvents are chosen using key performance indicators (KPIs) that quantify mass efficiency and product quality. The proposed methodology is illustrated by identifying promising reaction and crystallisation solvents for the synthesis of mefenamic acid from 2,3-dimethylaniline and 2-chlorobenzoic acid. Furthermore, multi-objective optimisation is deployed to highlight the trade-offs between the solvent or process E-factor and safety indicators, and between the solvent E-factor and crystal yield. The inclusion of mass-based KPIs and safety specifications ensures that only high-performance solvents are chosen. The findings of our approach are expected to guide the rational selection of solvents for greener pharmaceutical manufacturing during early-stage process development. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | None yet |
URL | https://zenodo.org/record/5555795 |
Title | Model-based solvent selection for integrated synthesis, crystallisation and isolation processes |
Description | We present a systematic process-wide solvent selection tool based on computer-aided mixture/blend design (CAMbD) (Gani, 2004) for the integrated synthesis, crystallisation and isolation of pharmaceutical compounds. The method proposed simultaneously identifies the solvent and/or antisolvent mixture, mixture composition and process temperatures that optimise one or more key performance indicators. Additionally, the method entails comprehensive design specifications for the integrated process, such as the miscibility of the synthesis, crystallisation and wash solvents. The design approach is illustrated by identifying optimal solvent mixtures for the synthesis, crystallisation and isolation of mefenamic acid. Furthermore, a multi-objective CAMbD problem is formulated and shows that a mefenamic acid with purity of 98.8% can be achieved without significant loss of process performance in terms of the solvent E-factor. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | Use in the development of pharmaceutical processes. New funding secured. |
URL | https://zenodo.org/record/5585546 |
Title | Model-based solvent selection for integrated synthesis, crystallisation and isolation processes |
Description | We present a systematic process-wide solvent selection tool based on computer-aided mixture/blend design (CAMbD) (Gani, 2004) for the integrated synthesis, crystallisation and isolation of pharmaceutical compounds. The method proposed simultaneously identifies the solvent and/or antisolvent mixture, mixture composition and process temperatures that optimise one or more key performance indicators. Additionally, the method entails comprehensive design specifications for the integrated process, such as the miscibility of the synthesis, crystallisation and wash solvents. The design approach is illustrated by identifying optimal solvent mixtures for the synthesis, crystallisation and isolation of mefenamic acid. Furthermore, a multi-objective CAMbD problem is formulated and shows that a mefenamic acid with purity of 98.8% can be achieved without significant loss of process performance in terms of the solvent E-factor. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | None yet |
URL | https://zenodo.org/record/5585545 |
Title | Modeling the thermodynamic properties of saturated lactones in non-ideal mixtures with the SAFT-? Mie approach. JCED 2023 |
Description | All computational data in the publication. |
Type Of Material | Database/Collection of data |
Year Produced | 2023 |
Provided To Others? | Yes |
URL | https://zenodo.org/record/8268755 |
Title | OMLT Optimisation and Machine Learning Toolkit |
Description | The optimization and machine learning toolkit (OMLT) is an open-source software package incorporating neural network and gradient-boosted tree surrogate models, which have been trained using machine learning, into larger optimization problems. OMLT seamlessly integrates with the algebraic modeling language Pyomo. |
Type Of Material | Computer model/algorithm |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | Higher-level representations, such as those available in ONNX, Keras, and PyTorch, are very useful for modelling neural networks and gradient-boosted trees. OMLT extends the usefulness of these representations to larger decision-making problems by automating the transformation of these pre-trained models into variables and constraints suitable for optimization solvers. |
URL | https://github.com/cog-imperial/OMLT |
Description | Imperial College / CCDC collaboration |
Organisation | Cambridge Crystallographic Data Centre |
Country | United Kingdom |
Sector | Academic/University |
PI Contribution | Providing direction, supervision, contributing to the generation of results |
Collaborator Contribution | Providing direction, supervision, contributing to the generation of results |
Impact | A methodology to identify intramolecular hydrogen bonding within molecular crystals. |
Start Year | 2020 |
Description | PharmaSEL |
Organisation | Eli Lilly & Company Ltd |
Country | United Kingdom |
Sector | Private |
PI Contribution | We are working closely with our partners as part of this Prosperity Partnership in order to transform medicines manufacturing for new modalities (e.g., peptides) as well as small molecules. |
Collaborator Contribution | Our partner is providing scientific input and sharing insights into the priorities and challenges faced by industry. |
Impact | A number of papers have been published. This collaboration includes chemical engineering and chemistry. |
Start Year | 2017 |
Title | AcRoPLS: Acurate Robust PLS |
Description | AcRoPLS (Acurate Robust PLS) provides a methodology to create accurate and robust PLS models based on solving a data-driven optimization problem that couples data pre-processing and model regression to a single optimization step. The accuracy objective is evaluated based on the performance of the generated model on predicting the model output on a test set. For the robustness objective, a novel metric based on the method of moments applied for different realizations of a known variability source evaluated again on a test set. |
Type Of Technology | Software |
Year Produced | 2023 |
Open Source License? | Yes |
Impact | Eli Lilly is discussing integrating this code into their tool chain. |
URL | https://github.com/ckappatou/AcRoPLS |
Title | CrystalPredictor 2 |
Description | Software for global search of crystal structures of organic molecules |
Type Of Technology | Software |
Year Produced | 2022 |
Impact | Use to reduce risk of finding unknown crystal structures; further funding |
Title | DEUS - DEsign under Uncertainty using Sampling techniques |
Description | DEsign under Uncertainty using Sampling techniques (DEUS for short) offers Nested Sampling based methods for solving the following problems: - Bayesian parameter estimation - Set-membership estimation - Design space characterization |
Type Of Technology | Software |
Year Produced | 2021 |
Open Source License? | Yes |
Impact | DEUS forms the basis for several publications on design space characterization in pharmaceutical manufacturing: https://doi.org/10.1021/acs.iecr.9b05006 https://doi.org/10.1016/B978-0-12-823377-1.50327-X https://doi.org/10.1016/j.ifacol.2020.12.555 https://doi.org/10.1016/j.ifacol.2021.08.222 |
Title | Pydex - Python Design of Experiments |
Description | An open-source Python package for optimal experiment design, essential to a modeller's toolbelt. Pydex helps design maximally informative experimental campaign for collecting data to calibrate a model mathematical, be it a mechanistic model or a statistical model. It is based on the concept of continuous efforts and relies on convex optimization. |
Type Of Technology | Software |
Year Produced | 2021 |
Open Source License? | Yes |
Impact | - Continuation of the project in collaboration with Siemens PSE for possible integration within the commercial process modeling and optimization platform gPROMS. |
Title | bootstrapPLS: Code for quantifying the uncertainty of a partial least squares prediction using bootstrap prediction |
Description | The software replicates the paper titled "Probabilistic predictions for partial least squares using bootstrap" (https://aiche.onlinelibrary.wiley.com/doi/full/10.1002/aic.18071) and develops an approach, based on bootstrapping, that automatically accounts for these nonlinearities in the parameter uncertainty, allowing us to equally well represent confidence intervals for points lying close to or far away from the latent space. The GitHub repository gives applications in determining the Design Space for industrial processes and modelling the uncertainty of spectroscopy data. |
Type Of Technology | Software |
Year Produced | 2023 |
Open Source License? | Yes |
Impact | We are discussing the possibility with Eli Lilly of integrating this into their tool chain. |
URL | https://github.com/jamesacodgers/bootstrapPLS |
Description | Great Exhibition Road Festival 2023 |
Form Of Engagement Activity | Participation in an open day or visit at my research institution |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Public/other audiences |
Results and Impact | Showcasing molecular design and explaining its importance to visitors of the Great Exhibition Road festival |
Year(s) Of Engagement Activity | 2023 |
Description | Imperial Festival |
Form Of Engagement Activity | Participation in an open day or visit at my research institution |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Public/other audiences |
Results and Impact | Crystal structure prediction is the field of science which aims at predicting what the crystal structure of a molecule will be in a solid. Crystal structure is very important in ensuring that medicines behave like we want them to and for metal to have desired properties. Tempering the chocolate alters the crystal structure of the chocolate to give it more desirable characteristics. PhD students demonstrated the importance of the "right" crystal structure of molecules, using chocolate as an example. Untampered or badly tempered chocolate and tempered chocolate were be on display for people to see which characteristics makes chocolate attractive and desirable and how tempering can help us in getting the desired crystal structure for the case of chocolates. A computer programme helped to recommend an optimal strategy for tempering the molecular characteristics. |
Year(s) Of Engagement Activity | 2022 |
URL | https://www.imperial.ac.uk/news/237506/great-exhibition-road-festival-2022-serves/ |
Description | Pharmaceutical Manufacturing Forum 2023 |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Professional Practitioners |
Results and Impact | A day of oral presentations, posters and panel discussions focussed on the future of pharmaceutical manufacturing. |
Year(s) Of Engagement Activity | 2023 |
Description | Research Team Visit to Eli Lilly & Co |
Form Of Engagement Activity | Participation in an open day or visit at my research institution |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Industry/Business |
Results and Impact | A team from UCL visited Eli Lilly in Indianapolis to engage with industry experts involved in pharmaceuticals manufacturing. |
Year(s) Of Engagement Activity | 2022 |
Description | Workshop on Model-Based Design of Experiments |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Postgraduate students |
Results and Impact | Industry practitioners, expert academics and PhD students attended this one day workshop to discuss the latest advances in Model-Based Design of Experiments. |
Year(s) Of Engagement Activity | 2023 |