DATA ANALYTICS THEME
Drupad Trivedi; Lucy Morgan
- Use of Machine Learning for Multiomics Data Integration
Background
This research addresses a key challenge in mass spectrometry data analytics: data integration. Omics approaches, which measure proteins, lipids, and metabolites in biofluids, microbial samples, or cellular systems, are commonly used to discover phenotypic markers of perturbation or processes of interest. While state-of-the-art analytical instruments are used to obtain measurements from biological or chemical systems, these data are often studied in isolation. Mass spectrometry (MS) is the most common analytical method used for multiomics, generating big data that are multidimensional and multimodal. High-resolution methods can introduce noise and often have missing chemical entities across different batches, leading to uncertainties in measurements. Existing open-source tools and wrappers for integrating multiomics data, particularly in biological systems, frequently suffer from poor data quality due to the nature of analytical sensitivity
Aims
- Develop advanced machine learning solutions to integrate and analyze multiomics data, particularly from mass spectrometry (MS), addressing issues like sparsity, heterogeneity, and multicollinearity.
- Improve the quality and usability of multiomics datasets for health research, process monitoring, and therapeutic development by addressing data uncertainty and incompleteness.
Objectives
- To create tailored algorithms using neural networks, tree methods, and probability-based algorithms for feature selection and biomarker identification from noisy and sparse MS data.
- To combine data from different omes (e.g., proteomics, lipidomics, metabolomics) to build predictive models that uncover patterns and relationships within the data.
- To test the developed solutions on publicly available datasets to ensure their effectiveness and reliability.
Expected Outcomes
- Enhanced methodologies for integrating multiomics data, leading to more accurate and comprehensive analyses.
- Development of robust predictive models that can identify biomarkers and phenotypic markers of perturbation or processes of interest.
- Generation of valuable insights for health research, process monitoring, and therapeutic development through the effective utilization of integrated multiomics datasets.
- Increased ability to utilize complex, multidimensional, and multimodal data from MS and other omics approaches, overcoming challenges related to data quality and heterogeneity.
2. Use of ML for High Fidelity Data ‘Stitching’
Background
Time domain and frequency domain data are two primary types of analytical measurements. Fourier transform is a common mathematical approach used to interchange data domains. A significant challenge lies in the meaningful integration of chemical information about the same system collected on different platforms, even within the same laboratory. For instance, analyses of an unknown wastewater sample using gas chromatography-mass spectrometry and ion-mobility mass spectrometry generate complementary information about the same sample. Typically, these datasets are treated as separate entities, leading to disjointed data reporting. Data stitching, a method used in other disciplines to acquire and aggregate data from different sources, can address this issue.
Aims
- Create machine learning solutions to integrate and analyze data from different analytical platforms, ensuring high fidelity in the combined datasets.
- Improve the quality and coherence of data reporting by integrating complementary information from different measurement techniques.
Objectives
- To investigate and adapt established methods from other disciplines, such as multiarray integration and batch correction, for high fidelity data stitching.
- To develop probability-based classifiers for robust feature selection, feature matching, and data aggregation.
- To test the developed techniques on datasets from multi-lab, multi-site studies to ensure their effectiveness and reliability.
Expected Outcomes
- Enhanced methodologies for integrating data from different analytical platforms, leading to more accurate and comprehensive analyses of biological or chemical systems.
- Development of robust techniques that ensure coherent and high-quality data reporting from integrated datasets when different analytical approaches are used.
Long Term Benefits
- Potential reduction in costs associated with running expensive standards by effectively utilizing data from multiple sources.
- Increased applicability of high-fidelity data stitching techniques in various fields, including health research, environmental monitoring, and industrial processes.
COMPLEX MIXTURES – SEPARATION & DETECTION THEME
Mansoor Saeed; Ali Salehi-Reyhani
Advancing Separation Technology: In the measurement of molecular species across different scales of molecular systems and complexity
Mission statement: Enhance and broaden separations science to encompass the separation and quantification of molecular systems across different length scales and complexity, from networks of molecules to molecular systems and machines.
Objectives
- Develop Novel Separation Processes:Explore and implement separation techniques beyond traditional chromatography, enabling efficient separation of components or entire molecular systems based on their chemical and biochemical properties across a broad range of species class, scale, and complexity. These can include complex cell lines, polymer mixtures, biotherapeutics (monoclonal antibodies, therapeutic proteins, fusion proteins, antibody-drug conjugates, and oligonucleotides), agrochemical biologics, and small molecules in agrichemicals and pharmaceuticals. To meet process and development requirements of the molecular species.
- Enhance Separation Flexibility:Design and implement separation systems adaptable to a wide range of molecular system scales (molecular size, number of species, complexity of the system and its mixture) and chemical characteristics, allowing for easy manipulation and method development across different scales of complexity, dependent on the molecular species and environment.
- Advance Detection and Characterization:Integrate cutting-edge detection technologies with high sensitivity and capacity for quantitative and qualitative characterization of molecular species, minimizing reliance on reference standards.
- Predictive Modelling:Develop computational models and simulations to accurately predict separation outcomes, enabling optimization and scalability of separation methods of the molecular species under investigation. Context aware platforms that can responsively adapt separation parameters.
Plan
Alternative Separation Processes
- Design, develop and evaluate novel separation techniques and platforms for their capability in measuring molecular species of concern in a complex mixture.
- Conduct experimental studies to assess the efficiency and effectiveness of the alternative separation process in separating components and molecular species based on molecular size, polarity, and chemical properties.
- Optimize and refine the selected alternative separation process for performance and applicability in industrial and academic settings.
Flexible Separation Systems
- Design and develop novel separation systems incorporating innovative materials and technologies (e.g., adaptive solid phases, separations in liquid phase without a solid phase, microfluidic devices) to adapt to a wide range of molecular sizes and chemical properties to meet the requirements of the molecular species.
- Explore multi-dimensional separations and hyphenated techniques beyond current traditional hyphenated separation modalities (e.g., LC-MS, CE-MS, IMS-MS) to extend the dynamic range and resolving power for a molecular species in a complex mixture.
- Implement modular and reconfigurable separation systems for easy manipulation and method development across diverse analytical needs.
Advanced Detection and Characterization
- Evaluate and implement state-of-the-art detection technologies by adapting or improving upon current detection technologies (e.g., High Resolution Accurate Mass Spectrometry, Ion Mobility Mass Spectrometry, advanced optical techniques) for sensitive and comprehensive characterization of a molecular species in a complex mixture.
- Develop and validate analytical methods for accurate quantitation and qualitative analysis of molecular species in multicomponent mixtures, minimizing reliance on reference standards.
- Integrate data processing and analysis tools (chemometrics, machine learning) for efficient interpretation and visualization of complex data sets.
Predictive Modelling and Simulation
- Develop computational models and simulations that accurately predict separation outcomes based on physicochemical molecular species properties, separation conditions, and system parameters.
- Validate and refine the models through experimental data for accurate prediction and optimization of separation parameters to meet the requirements of the molecular species under investigation.
- Integrate artificial intelligence techniques (deep learning, reinforcement learning) to enhance predictive capabilities and accelerate method development.
Expected Outcome
The expected outcome is the development of innovative separation methodologies, adaptable separation systems, advanced detection and characterization techniques, and predictive modelling tools. These advancements will collectively bolster the capabilities for tackling multicomponent analysis in intricate mixtures across a broad range of molecular species, including complex polymer mixtures, biotherapeutics, agrochemical biologics, and small molecules in agrichemicals and pharmaceuticals.
Bringing together new disciplines and taking a molecular systems engineering approach across length scales and molecular complexity will help broaden the scope and ambition of separation sciences as our chemistries develop beyond small molecules, to biomolecules and coordinated chemical machines.
The ultimate goal is to establish separation and detection platforms that minimize reliance on reference standards, offering a comprehensive solution for the analysis of diverse molecular species and spectrum of applications and scales of complexity, while going beyond traditional separation modalities to meet the 21st century separation requirements.
SENSORS AND PHOTONICS THEME
Binoy Paulose; Natalie Belsey
1) Photonics based solutions for in-line/in-vivo monitoring of processes:
Photonics-based solutions are increasingly being used for in-line and in-vivo monitoring of various processes due to their precision, non-invasiveness, and ability to provide real-time data. These solutions utilize light-based technologies, such as lasers, optical fibers, and sensors, to monitor chemical, biological, and physical changes within a system without direct contact or disruption. In industrial applications, photonics can be used to monitor manufacturing processes, ensuring product quality and consistency. In medical and biological contexts, photonics-based techniques enable the monitoring of physiological parameters and disease markers inside the body, offering a minimally invasive approach that can lead to improved diagnostics and patient outcomes.
Aims:
This challenge focusses on developing photonics-based solutions for in-line and in-vivo monitoring of processes across various fields including manufacturing and healthcare. In manufacturing, these solutions can be aimed at enabling non-destructive and real-time monitoring, facilitating precise control and optimization of processes by continuous monitoring of process parameters including chemical composition, temperature, and pressure within production lines with high sensitivity to ensure product quality and safety while minimizing waste and production downtime.
In healthcare, this challenge can be aimed at developing tools for non-invasive in-vivo monitoring of biological processes within tissues and organs in real-time, allowing clinicians to monitor physiological changes without invasive procedures to aid diagnosis, treatment monitoring, and personalized medicine, ultimately enhancing patient care and outcomes.
Objectives:
- Development of photonics-based solutions for accurate and reliable measurements in various environments, from biological tissues or to industrial manufacturing sites, where challenging conditions include movement and vibrations, temperature, and complex sampling composition can affect light propagation and data accuracy.
- Development of compact, high-performance photonic devices suitable for in-vivo applications, which requires advancements in miniaturization while maintaining measurement capability, integration with existing systems, and overcoming limitations in size, weight, and power consumption.
- Development of affordable and scalable photonic devices for widespread adoption in both medical and industrial applications.
2) Point of Use Sensors for quantitative biological measurements
Point of use sensors for quantitative biological measurements are designed to provide rapid, accurate, and on-site analysis of various biological parameters directly at the location where they are needed. These sensors are often portable and user-friendly, enabling non-specialist users to obtain critical information quickly without the need for a laboratory setting. They utilize various detection methods, such as electrochemical, optical, or piezoelectric transducers, to measure specific biological markers. The ability to perform real-time, quantitative measurements at the point of use is particularly valuable in clinical diagnostics, environmental monitoring, and food safety, as it allows for timely decision-making and intervention. These sensors contribute to personalized medicine and decentralized testing, improving overall healthcare outcomes by facilitating more frequent and accessible monitoring of health conditions.
Aim:
This challenge aims to develop point-of-use sensors for quantitative biological measurements to enable rapid and accurate detection of biomolecules such as proteins, nucleic acids, and metabolites at the point of use.
The goal of the challenge is to advance sensor design, multiplexing capabilities, detection methods and their integration to address complex biological challenges to improve quality of life.
Objectives
- Developing point-of-use sensors for improved sensitivity to detect low concentrations of target biomolecules and high specificity to distinguish these targets from similar, non-target molecules in complex biological samples.
- Developing stable and reliable point-of-use sensors for better performance over time, especially when exposed to varying environmental conditions such as changes in temperature, humidity, and chemical exposure, without compromising their quality, accuracy, or durability.
- Developing point-of-use sensors that are compact, portable, and easy to integrate with other devices or systems, while maintaining their accuracy and precision in measurements.
- Development of intuitive, easy-to-use interfaces and data interpretation tools that allow non-specialist users to operate the sensors effectively and understand the results accurately without extensive training.
NOVEL INSTRUMENTATION THEME
Michael Wilde; Dara Fitzpatrick
1 . Development, Evaluation and Application of Next Generation, Replacement or Novel Instrumentation for Sustainable (Green) Measurements.
This challenge is aimed at developing the next generation of analytical measurement technologies for laboratory, field or production environments which minimise environmental impact (including consumables and waste). Proposals could include (but are not limited to)
Next generation: e.g. miniaturisation of existing technologies.
Replacement (i.e. existing technologies) that are ‘greener’ than typical / industry/regulatory accepted approaches but which need full evaluation and demonstration of analytical ‘equivalence’.
Novel technologies which have excellent ‘green’ credentials – offering significant advantages in minimising environmental impact as well as in measurement capability.
2 . Introduction of novel Measurement Technologies
This challenge is aimed at novel measurement technology which offers distinct advantages over existing techniques or which allows measurement of components previously challenging or impossible to measure by existing techniques. The technology may have already been developed but needs further evaluation for (new) industry applications or the technology itself may need further development and refinement. It may be that the instrumentation has yet to be developed but the proof of concept and or theory strongly supports the potential success. Lab and miniaturisation scales are in scope for this challenge.
