Yu-Yuan (Stuart) Yang

Profile

Project Title: Machine learning for structure-based drug & biologic discovery 

Summary: Rational approaches to drug discovery heavily rely on the availability of suitable 3D structures for the target binding sites. Using a single structure, even if highly accurate, can lead to very misleading results if the binding site is not in a conformation able to bind a ligand. Owing to the inherent plasticity of receptors, ensemble docking is harnessed to augment the accuracy of docking predictions by employing multiple conformations extracted from molecular dynamics (MD) trajectories as representative models. However, this is associated with an increased computational cost compared to single-structure approaches.

As such, the aim of this study is to develop a machine learning-based tool to identify structures that are most representative of the different shapes adopted by binding sites in MD simulations, serving to mitigate the computational costs of structure-based drug discovery approaches while maintaining requisite accuracy. Anticipating its extensive applicability, this tool is poised for further development into a user-friendly graphical interface, thereby enabling users to efficiently analyze and visualize their data.

The developed methodology will then be applied to explore ensemble dynamics for biologics drug discovery with a focus on single-chain antibodies and RNA aptamers.

Supervisors:

• Prof Richard Pickersgill 
• Dr Arianna Fornili 

Research