Industry Challenges

 

GSK  

High Throughput Automated Imaging and Processing of Solid/Liquid Samples For Pharmaceutical Development

CAMS themes:

• Novel instrumentation or technique
• Data analytics
• Complex mixtures, separations and detection

Project Summary

In development of novel pharmaceutical compounds and medicines, pharmaceutical companies and associated industrial partners regularly conduct the determination of solubility of drug substances, drug substance intermediates, and pharmaceutical excipients in both organic and aqueous solvents. The solubility data is then subsequently employed in a myriad of processes including for example analytical method development, formulation development, chemical process development and Physiologically based pharmacokinetic (PBPK) modelling. Quantitative measurements of solubility can be conducted either in real-time using suitable process analytical technologies or by sampling and subsequent quantitative analytical techniques. In addition to quantitative data, qualitative data is often gathered via visual observation and imaging.  This may include capture of, for example, clarity of sample, colour change, formation of agglomerates, wetting of solids, formation of multiphase systems etc. Obtaining a data set of observations through these processes can be time-consuming and the resulting data set complex. When automated analysis and automated solid dispensing and liquid handling systems are employed, then many hundreds of qualitative observations may be generated concurrently to quantitative measurements. In these instances, processing and interpretation of observations via Humans may lead to a) inconsistency / ambiguity in description and interpretation b) large amounts of time spent in conducting observations and c) physical intervention being required in an automated workstream. GSK would like to discuss the potential of leveraging the CAMS network for academic / industry collaboration to research the application of image collection, image processing, data collation and interpretation of qualitative observations in a fully automated, time sensitive, High Throughput manner.             

Additional Comments

Contributors with expertise in the following:

• Robotics
• Imaging
• Machine Learning
• Liquid Handling
• Data Handling

Could be possible to break research into multiple parallel complimentary projects

 

GSK  

Development of SERS-based Sensors for Oligonucleotide Drug Investigation

CAMS Themes

Novel instrumentation or techniques

Project Summary

This groundbreaking research project seeks to pioneer the use of Surface Enhanced Raman Spectroscopy (SERS) sensors in the pharmaceutical industry, with a focus on oligonucleotide drugs. The project will develop cutting-edge sensors capable of detecting precise structural changes, such as high-order structures and base mismatches, that are critical in drug development and quality control. By leveraging the unparalleled sensitivity and specificity of SERS, this study aims to provide unprecedented insights into the molecular dynamics of nucleic acid-based therapies. This innovation holds the promise of transforming the monitoring and quality assurance processes in drug manufacturing, setting a new standard for the development of safe and effective pharmaceutical products.

Milestones and Deliverables:

• Literature Review and Theoretical Framework
• Comprehensive review of current SERS applications in nucleic acid research.
• Establishment of a theoretical framework for SERS-based detection of oligonucleotide drugs.
• Sensor Development and Optimization
• Fabrication of SERS-active substrates using silver/gold colloids.
• Optimization of substrate preparation for enhanced sensitivity and reproducibility.
• Experimental Design and Data Collection
• Systematic study of SERS spectra of single-stranded and double-stranded oligonucleotides with controlled base sequences and modifications.
• Investigation of environmental effects (pH, ionic strength, temperature) on SERS signals.
• Data Analysis and Interpretation
• Use of multivariate analysis to interpret SERS spectra and identify specific molecular interactions.
• Correlation of spectral data with structural changes and environmental conditions.
• Application and Validation
• Testing of SERS sensors on real-world samples, including oligonucleotide drugs.
• Validation of the sensor's accuracy and reliability in detecting base mismatches and structural anomalies.

 

Additional comments

Summary of Potential Project Benefits

Enhanced Detection Sensitivity: SERS sensors offer unparalleled sensitivity due to their ability to enhance Raman scattering signals by several orders of magnitude. This sensitivity allows for the detection of minor structural changes, such as single nucleotide polymorphisms (SNPs), methylation patterns, and other modifications. These minute changes can significantly impact the function and efficacy of oligonucleotide-based drugs. The high-resolution spectra obtained from SERS can differentiate between closely related molecular structures, making it an invaluable tool for quality control in the pharmaceutical industry, ensuring that the final product is consistent and free of undesired modifications.

Broad Application Range: The versatility of SERS technology extends beyond simple detection, enabling its application in a wide array of nucleic acid-based therapies. In gene therapy, SERS can be used to monitor the incorporation of therapeutic genes and the expression levels of targeted sequences. For antisense therapies, which involve the use of oligonucleotides to block the expression of specific genes, SERS can help in evaluating the binding efficiency and specificity of the antisense molecules. Additionally, the technology can be adapted for use in CRISPR/Cas9 applications, providing real-time monitoring of gene editing processes and ensuring precision in genetic modifications.

Real-Time Monitoring: One of the significant advantages of SERS sensors is their capability for real-time monitoring of molecular interactions. This feature is particularly beneficial in studying oligonucleotide interactions in various environments, such as different pH levels, ionic strengths, and temperatures. By continuously monitoring these interactions, SERS can provide insights into the stability and dynamics of oligonucleotide structures under physiological conditions. This real-time data is crucial for understanding how environmental factors influence the efficacy and stability of therapeutic oligonucleotides, guiding the design of more stable and effective drug formulations. Furthermore, the ability to perform in situ monitoring without extensive sample preparation makes SERS an attractive option for clinical diagnostics and personalized medicine, where rapid and accurate analysis is essential.

Novelty and Scope for Publication

The proposed research is novel in its comprehensive approach to studying oligonucleotide drugs using SERS. Currently, there are no SERS-based sensors specifically tailored for the pharmaceutical industry to monitor drug development processes, particularly for oligonucleotide-based therapies. This project aims to fill this gap by developing sensors capable of detecting high-order structures, base mismatches, and other critical modifications with high sensitivity. By focusing on these aspects, this project has the potential to uncover new insights into the stability and functionality of nucleic acid drugs. The development of such sensors could revolutionize quality control and monitoring processes in drug manufacturing, providing a novel tool that ensures the production of consistent and effective pharmaceuticals. The findings and methodologies could lead to several high-impact publications in journals specializing in analytical chemistry, nanotechnology, and pharmaceutical sciences.

Probability of Technical Success

Given the advancements in SERS technology and previous success in detecting nucleic acid sequences, the probability of technical success is high. Challenges such as signal reproducibility and substrate preparation will be addressed through rigorous optimization and validation processes. Additionally, this project will collaborate with leading expert in SERS to provide invaluable insights and support, ensuring the project's success.

An expanded project proposal is available from Adam Watkins (please contact via CAMS Secretariat).

 

GSK  

Studying the application of state-of-the-art machine learning approaches to predict the crystal structures of polymorphs

CAMS Themes

• Novel instrumentation or techniques
• Data analytics

Project summary

The proposed PhD project will develop crystal structure prediction methods using the state-of-art machine learning technology of MACE-OFF23, a transferrable atomistic forcefield for organic molecules. The use of this forcefield will be explored in predicting the minima energy polymorphic structures of small molecule active pharmaceutical ingredients (APIs) compared to other ab initio based approaches. In addition, predicted structures that are generated using MACE-OFF23 models will be compared to analytical experimental data to elucidate the correct methodology in using the MACE architecture for theoretical structure prediction. Finally, the scalability and applicability of MACE-OFF23 in predicting crystal structures of large systems such as amorphous APIs or proteins can be assessed.

Milestones and deliverables

1. October 2025 – July 2026 (9 months) Implementation and exploration of MACE-OFF23 methods – predicting the structure of a known polymorphic structure.
2. July 2026 – October 2026 (3 months) Completing ab initio based approach for comparison
3. October 2026 – April 2027 (6 months) Application of methodology to compound with an unknown polymorphic structure.
4. April 2027 – Oct 2027 (6 months) Acquisition of analytical experimental data to be used as confirmation of crystal structure prediction of unknown polymorphic form.

5.October 2027 – October 2028 (12 months) Looking at the application to larger systems i.e. amorphous structures.

Additional Comments

Summary of potential project benefits

The proposed crystal structure prediction project will mostly benefit the analytical chemistry and materials science communities as prediction of minima energy structures will aid the understanding of drug physicochemical properties.

Knowledge of the crystal structure of an active pharmaceutical ingredient is essential in the development of a drug for several reasons:

- Polymorphism: Each polymorph of a drug can have different physical and chemical properties; therefore, knowledge of different polymorphs enables an understanding of the relationship between drug physicochemical properties and stability, safety, and efficacy.

- Solubility and Dissolution:  The solubility and dissolution drugs in development can significantly vary between different polymorphic forms of an API which is a critical factor in determining a drug’s bioavailability.

- Stability: Different crystal forms will have varying stability profiles and knowledge of polymorphic forms can aid the identification of the most stable form to develop.

- Regulatory approval: An understanding of the crystal structure helps in meeting regulatory agencies requirements in ensuring a drugs quality, safety, efficacy and manufacturability.

The primary routes in determining a crystal structure analytically are through diffraction-based methods (X-ray, electron, or neutron diffraction) which provide accurate representation of atomic positions. However, numerous challenges can arise in experimentally generating a crystal structure, such as difficulty in obtaining a suitable single crystal of the drug, not being able to synthesise phase-pure material or the lack of stability of a polymorph. This often means it is not possible to create a representative accurate crystal structure of all polymorphic forms and a significant amount of time and resource may be spent in creating structures which may not be fully representative of the true crystal structure. Crystal structure prediction is being used more and more as a tool to solve such issues where experimental analytical data is used as guidance as to whether the correct crystal structure has been predicted.

Previously, the primary route for materials modelling has been to use ab initio density functional theory (DFT) models.[1] The high computational cost of such calculations has meant that there are limitations on the size and accuracy of the system which can be modelled where increasing the size of the system can come at the cost of generating an inaccurate model. This has limited crystal structure prediction using DFT models to structures which contain only a few atoms, with any attempt to apply this to larger organic molecules becoming inaccurate and too computationally heavy, which has previously often been the case with pharmaceuticals. The atomistic modelling of materials has been revolutionised in recent years through the generation of machine-learned force fields.[2,3] The further recent advances of artificial intelligence and machine learning have allowed for improvements in the accuracy, robustness and computational speed in generating such force fields for inorganic crystals through the use of the state-of-the-art machine learning MACE architecture.[4] Recently, a transferable force field for organic molecules has been based created based on this architecture, called MACE-OFF23. [5] This has a linear scaling in computational requirement meaning that comparatively to other force fields it can scale up to model systems as large as proteins without the cost of sacrificing accuracy for speed. From initial studies using MACE-OFF23 it has been possible to simulate a fully solvated small protein, reproduce the experimental properties of condensed phase molecular systems, and accurately predict the Raman spectrum of the “Form II” polymorph of paracetamol.[5]

An extended version of this proposal is available from Adam Watkins (Contact Via CAMS secretariat).

 

GSK  

Studying the application and implementation of NQR in pharmaceutical development       

CAMS Themes

• Data analytics
• Novel instrumentation or techniques

Project Summary

The proposed PhD project will develop nuclear quadrupolar resonance (NQR) spectroscopy methods to look at the structural characterisation and quantification of polymorphic forms. This project will focus on the use of 35/37Cl and 14N NQR and its application to a few key areas in pharmaceutical development such as detection and quantification of mixtures of polymorphic forms, the analysis of hydrogen bonding, tautomerism and the characterisation of crystal structure defects and impurities. Furthermore, a significant challenge associated with NQR is the difficulty in predicting the frequency at which a signal will appear. Therefore, a primary focus of this project will also be to develop methods to theoretically determine quadrupolar frequencies of 35/37Cl and 14N nuclei to allow for NQR to become a more routinely used technique.

Milestones and deliverables

1. October 2025 – April 2026 (6 months) Detection of polymorphs and compounds with known quadrupolar moments to develop a setup for NQR.
2. April 2026 – April 2027 (12 months) Development of a predictive method using possibly using density functional theory calculations or creating databases for determining quadrupolar frequencies.
3. April 2027 – October 2027 (6 months) Investigate the quantification of mixtures of polymorphic structures and determine detection and quantification levels achievable by 35Cl NQR.
4. October 2027 – June 2027 (9 months) Characterisation of hydrogen bonding and tautomerism present within the structure using NQR.
5. June 2027 – October 2028 (3 months) Further applications, potential areas may include the detection of amorphous materials, phase transitions and crystallisation processes using variable temperature.

 

An extended version of this project proposal is available from Adam Watkins (Contact Via CAMS secretariat).

Additional Comments

The proposed project will mostly benefit the analytical chemistry and materials science communities as the introduction and implementation of this technique will allow for a new approach in quantification and structural characterisation of pharmaceuticals.

NQR is a spectroscopic technique which is analogous to Nuclear Magnetic Resonance (NMR), but with some key differences. In a typical NMR experiment a strong magnetic field is applied which causes a Zeeman splitting of the atomic energy levels to occur. This happens for all nuclei that have the intrinsic property of spin (I>0), allowing for transitions between these energy levels to be observed through excitation with RF pulses. NQR on the other hand does not require a strong magnetic field which therefore makes the technique much cheaper to facilitate and run. Instead of using an external magnetic field, NQR relies on the energy level splitting caused naturally by quadrupolar nuclei (I > ½) where the quadrupolar interaction perturbs the energy levels of nuclear spin-states. Similarly to NMR, the perturbations of different nuclei can be observed allowing for valuable information to be extracted about the crystal structure. Quadrupolar nuclei (I > ½) observed by solid state NMR often have enormous convoluted and complex lineshapes for organic compounds making these nuclei unfeasible in the characterisation of pharmaceuticals. [1] However, when using 35Cl or 14N NQR simple gaussian peaks are observed which are easily analysed and will be acquired in significantly less time when compared the acquisition of these nuclei by solid state NMR. [2]

Currently, the primary limitation of NQR and the reason for the lack of development when compared to solid state (SS) NMR is the need to ‘search’ over wide spectral windows to locate a resonance. For example, a covalently bonded 35Cl nucleus will have a resonance which can emerge over a ~15 MHz range. This becomes a challenge with limitations in experimental equipment typically only able to excite a region in the hundreds of kHz meaning searching the entire 35Cl region can be very time consuming. There are currently methods which make it possible to predict the quadrupolar moments of these quadrupolar nuclei primarily in studies for solid state NMR. [3] This project will look to implement similar methods to predict the quadrupolar frequencies and therefore minimise the time taken to locate NQR resonances. The project will look to predict the resonances of organic pharmaceutical compounds to within 1 or 2 MHz either through density functional theory (DFT) calculations or through generating a database of quadrupolar couplings which will vastly improve the time taken to locate resonances. This will in turn allow NQR to become readily usable to analyse pharmaceuticals.

With the ability to locate resonances faster, quadrupolar nuclei such as 35Cl which are not feasible options using SSNMR can be analysed and can provide a greater sensitivity to structural changes than typical spin-1/2 13C, 19F nuclei observed by SSNMR. NQR has shown to be highly sensitive in the investigation of different polymorphs, tablet compaction and crystal structures. [4] However, there has been little exploration into the use of NQR to quantify different polymorphic forms. Quantification of polymorphic forms is frequently needed within the pharmaceutical industry as control measures in drug development. This is primarily done using X-ray diffraction, however, there are cases where solid state NMR is a better option for quantification. For solid state NMR, accurate and sensitive quantification is usually performed through highly sensitive spin-1/2 nuclei such as 19F or 31P. When 19F or 31P are not present in the target molecule 13C is used but the lower natural abundance and receptivity of the 13C nucleus makes quantification challenging. If the project was successful then this would allow for 35Cl and possible 14N to be used in the quantification of polymorphic forms.

 

Shimadzu Research Laboratory     

Developing diode laser-based technologies to enhance molecular characterization by tandem mass spectrometry  

CAMS Themes

• Novel instrumentation or techniques
• Point of use sensors and photonics

Project Summary

Mass Spectrometry (MS) is a major analytical measurement technology, pivotal across academia, government/institutions and industry. It is critical to any well-founded analytical laboratory and is of strategic importance in all aspects of molecular science. Central to this is tandem MS, the ability to break a molecule apart to deduce structure, concentration and/or activity/reactivity. Photodissociation (PD) uses photon energy to induce fragmentation and perfectly complements other, more traditional and common-place fragmentation techniques. The result is significant improvement in the depth of analytical information obtainable.

The aim of this project is to develop advanced diode laser-based technologies aimed at enhancing molecular characterization capabilities of tandem mass spectrometry (MS/MS). Diode lasers are cost-effective, easy to operate, and require no alignment, and offer advantages such as narrow bandwidth, tunability, and spatial coherence, making them ideal for applications in spectroscopy and mass spectrometry. By integrating these lasers into MS/MS systems, we aim to make this technology accessible, and to improve the breadth of molecular analysis, with potential applications in biomedical research, environmental monitoring, and material science.

Additional Comments

This project idea has been developed in collaborative discussions with Dr Jackie Mosely and Prof Caroline Dessent of the University of York.

 

Syngenta

Live assay High-resolution mass spec analysis

CAMS Themes

• Complex Mixtures Separation and Detection
• Novel Instrumentation/Techniques

Requirements

• Generation of small molecule high-resolution mass spectrometry data from Biological samples
• Assays samples are live in-vitro assays and benefit from continuous sampling for semi-continuous kinetic data

Additional Comments

• Could utilise high-throughput sample handling instrumentation (e.g. Echo, Rapidfire), a continuous ambient ionisation approach, or a more traditional LC-MS approach
• High-res mass spec is important due to the capturing of data for xenobiotic metabolite identification from these in-vitro ADME assays
• Potential technique also has applications for chemical/photochemical in-vitro system monitoring, and live Biochemistry IC50/SAR data generation.

 

Syngenta

Explore the combination of Electron Diffraction with the Crystalline Sponge Method to determine chemical structures of low-level impurities and metabolites from matrices.

CAMS Themes

• Complex Mixtures Separation and Detection
• Novel Instrumentation/Techniques
• Data Analytics

Requirements

• Straightforward, adaptable protocol that covers a broad spectrum (size and polarity) of possible analytes.
• Works with small, partially purified samples

Additional Comments

• By using electron diffraction, you can work with smaller crystals.
• This opens up the crystalline sponge space – usually a Metal Organic Framework (MOF).
• The goal is to get selective uptake of desired compounds using designed MOFs.  
• This is all highly speculative at this stage because there is only limited data on the application of electron diffraction to MOFs and even less information on how to absorb compounds selectively into the sponges.

 

Syngenta

Understanding fermentation process in cows

CAMS Themes

• Complex Mixtures Separation and Detection
• Novel Instrumentation/Techniques
• Data Analytics
• Point of use

Requirements

• Investigation of corn silage fermentation processes, ruminant biochemistry and detection methods that evaluate value added effects of new products.

Additional Comments

• New area of R&D which is not well understood.
• Determining feasilbility phase.
• Innovative solutions required to study ruminant. fermentation processes and how exdogenous compounds affect the endogenous profile.

 

Syngenta

Direct detection of Active Ingredient Volatilization and Movement (as gas or aerosol) AFTER its application.$Complex Mixtures Separation and Detection 

CAMS Themes

• Novel Instrumentation/Techniques
• Data Analytics
• Point of use

Requirements

• Direct detection of low-level active ingredients in the gas phase (either as gas, or aerosol particle)
• Building understanding, and characterisation portfolio to study molecular vapour /molecular diffusion of volatile-liquid vapours into air
• Ideally technique can be used to quantitate amount of active ingredient, or at least allow relative comparisons

Additional Comments

• Some AI show presence outside the area of application after application which is regulatory issue. We want to improve our understanding of processes driving movement of the AI in the atmosphere.
• We need to understand how much of AI is escaping from the surface (of soil, plant). Does the AI evaporate, volatilise, or dissolve in water vapour and is taken up to the atmosphere.
• The standard methods, comprises an indirect detection using extraction from a substrate. For example, glass slide in wind tunnel, pulling gas through a filter in laboratory experiments. We want to have DIRECT detection – ideally quantitative direct detection.
• We are interested in evaluating optical spectroscopy / laser techniques such as cavity ring-down and photoacoustic spectroscopy to study spray drift as well as molecular vapour and vapour diffusion. However, we are also open to hear about alternative approaches.

 

Syngenta

Stability of compounds in DMSO solubilised samples for long term storage in Central Research Dispensaries

CAMS Themes

• Data Analytics
• Point of use

Requirements

To develop a predictive algorithm that will assess compound stability in DMSO and other solvents. We routinely use LC/MS to determine compound stability in DMSO solubilised samples that are stored for 2-20 years in the Central Research Dispensary. The idea here is to feed the machine learning algorithm with LC/MS data results for Syngenta compounds. A model can then be generated to predict chemical moieties that are prone to degradation or other instabilities based on chemical structure and real analytical data.

Additional Comments

Expertise required:

• Computational Chemistry
• Extensive knowledge in machine learning algorithms
• Knowledge of processing and interpreting LC/MS, GC/MS and NMR datasets
• An understanding of the use of central research dispensaries in industry

 

Syngenta

Streamlining metabolite annotations from untargeted metabolomics LC-MS/MS datasets

CAMS Themes

• Data Analytics

Requirements

• Use state-of-the art structure mining and MS/MS spectral mining tools to annotate untargeted metabolomics data.
• Enable the quick annotation of batches of MS/MS spectra from one or several samples.

Additional Comments

• We want to test and optimise structural mining tools to annotate MS/MS spectra whose reference MS/MS spectra might not exist in spectral databases,
• In addition, we also want to leverage the use of spectral similarity tools for structural elucidation of unknown compounds, particularly natural products,
• Finally, we want to develop a pipeline for the batch annotation of MS/MS spectra using the methods mentioned above.

 

Syngenta

Rapid detection and quantification of bacteria within product formulations.

CAMS Themes

• Novel Instrumentation/Technique
• Complex Mixtures Separation and Detection

Requirements

• Syngenta is seeking solutions that allow the rapid detection and quantification of bacteria within their product formulations.
• Bacterial contamination can lead to compromised product, bulging or ruptured packaging and subsequent product recalls and waste. Therefore, it is key that proposed solutions can both detect and quantify the amount of bacterial contamination within a product so that appropriate action can be taken.

Additional Comments

• Current methods are time consuming (1-4 days) and require very specific technical expertise. Therefore, solutions that can simplify and speed up this process are of interest.
• Ideally the solution will allow this process to be performed easily and quickly within the manufacturing environment. Solutions will need to be able to detect bacteria at low concentrations.