(EPSRC CASE) Elevating Matched Molecular Pair Analysis with AI

About the Project

Discovering and designing new medicines is one of most important and rewarding challenges faced by chemical scientists. Maximising the impact of computational tools to improve the speed of the drug discovery process and the quality of the compounds that are invented has great potential to improve human health and wellbeing. One of the tools that has been most successful against this challenge is matched molecular pair analysis. Using cutting-edge chemical informatics tools to find pairs of molecules that share structural transformations allows predictions to be made about what the effect of that transformation will be in future uses. The best structural transformations for any given need can be identified quickly and prioritised. Developing the approach further, Medchemica (the CASE partner for this project) have developed a number of other tools that use matched molecular pair analysis. One example is their SARkush tool which can summarise what is already known about a chemical series in a representation that is very similar to Markush structures that are used in patents to describe the chemical scope of an invention.

AI has impacted in many areas of life, including in the drug design field. In this project you will explore the potential for marrying matched molecular pair analysis with AI. Two specific projects that will be at the heart of this PhD will be 1) providing new tools for naming and classifying parts of chemical structures and 2) exploring the potential of generative AI to suggest and rank new structural changes. In the first project, you will use socialised approaches in which human input will be used to train and scrutinise AI-based models. The objective will be to develop algorithms that can be provided with the structure of a part of a molecule (e.g. a methyl group) and be able to reliably provide a name for that part of a molecule and to be able to classify it e.g. as a small alkyl group. In this way, you will transform the ability of the SARkush system and other approaches to provide meaningful output to users. You will also create tools that will enable chemists and patent law experts to draft improved patents. These naming tools will be made available to the wider chemical community. In the second project you will work with encodings of structural changes called SMIRKS and explore the ability of generative AI approaches to accurately and reliably envisage previously unseen structural changes and predict their effects.

You will join a friendly group, based in the division of pharmacy and optometry, that will provide you with many opportunities to work with a wide range of scientists. You will undertake a placement with Medchemica where you will experience industrially relevant drug design.

Entry Requirements

Applicants should hold (or be about to obtain) a First or Upper Second class (2:1) UK honours degree, or international equivalent, in a relevant subject.

Application Guidance

Candidates must contact the primary supervisor before applying to discuss their interest in the project and assess their suitability.

Apply directly via this link: OAA Applicant Portal or on the online application portal, select MRC DTP PhD as the programme of study.

You may apply for up to two projects within this scheme. To do so, submit a single online application listing both project titles and the names of both main supervisors in the relevant sections.

Please ensure that your application includes all required supporting documents: Curriculum Vitae (CV), Supporting Statement, Academic Certificates and Transcripts.

Incomplete or late applications will not be considered. Further details are available on our website: EPSRC Doctoral Training Partnership | Biology, Medicine and Health | The University of Manchester