Expanding the Chemical Universe: 3D Features Driving Next-Gen Synthesis Predictions

Description

Background: Small molecule drug discovery projects are time-consuming and expensive. More than half of that cost and even more of the time of each project is taken up by chemical synthesis. This is in part driven by our inability to predict which reactions will work and which synthesis routes are most efficient. Therefore, forward synthesis prediction models are crucial for improving the efficiency of drug discovery and chemical synthesis. While these models have become increasingly accurate, they often focus on a limited set of reactions, struggling to generalize beyond well-established chemical space. Critically, current models do not offer any meaningful enhancement over a human chemist and worse still, they constrain computational design to a very limited segment of chemical space. Recent advancements have integrated 3D¹ and quantum mechanical (QM)² features into these and other models³, leading to improved predictions for specific reactions and broader forward synthesis prediction. However, it remains unclear which 3D features provide the most significant benefits and in what contexts they should be applied. 

Project Aim: This PhD project aims to systematically investigate the role of 3D conformer and descriptor information in enhancing the generalizability of forward synthesis prediction models. The goal is to determine how and when to incorporate 3D information to optimize model performance. This project will also explore the development of novel generative approaches⁴ that effectively utilize this information to enhance compound design.

Expected Outcomes

This project is expected to deliver:

• A comprehensive understanding of the role of 3D information in forward synthesis prediction models and its impact on model generalizability.
• Practical recommendations for incorporating 3D features into these models to improve their ability to extrapolate.
• Validated models with enhanced predictive capabilities, supported by new synthesis data.

Ideal Candidate

The ideal candidate will have a background in computer science or chemistry with a desire to learn the other domain. Experience with coding and ambition to learn more will be advantageous as will 3D molecular modelling, synthesis prediction, and data analysis, but are not essential. This project offers a unique opportunity to contribute to the development of next-generation tools in chemical synthesis prediction and apply them in real-world chemical synthesis.

How to Apply Formal applications should be made to How to apply for a PhD programme - University of Liverpool quoting reference CCPR132.Candidates will be evaluated as applications are received, and the position may be filled before the deadline if a suitable candidate is identified.

Please ensure you include the project title and reference number CCPR132 when applying.

Supervisors

Anthony Bradley - anthony.bradley@liverpool.ac.uk

John Ward - john.ward@liverpool.ac.uk

Fergus Imrie - fergus.imrie@stats.ox.ac.uk