Research Topic

Data-Driven Ligand Optimization

Traditional approaches to ligand design involve synthesising numerous structural variants — a process that is time-consuming and expensive. We aim to leverage machine learning to make this process dramatically more efficient.

Asymmetric Catalysis Graph Neural Networks Explainable AI Cheminformatics

Research Overview

Machine Learning for Catalysis

My research leverages Graph Neural Networks (GNNs) to revolutionise the optimisation of chiral ligands in asymmetric catalysis. By integrating data-driven approaches, my work accelerates the traditional ligand design cycle — reducing both the cost and effort required to achieve high enantioselectivity.

HCat-GNet: Interpretable AI

I focus on the development of interpretable AI models such as HCat-GNet (Homogeneous Catalyst Graph Neural Network), which predict catalytic reaction outcomes from molecular structure. This approach not only enables rapid ligand optimisation but also provides chemists with actionable insights into how molecular features influence reaction selectivity.

Broader Impact

This research has broad implications for the catalysis community, offering a faster and more accurate pathway to catalyst discovery. By automating the design process with interpretable tools, I aim to drive innovation in sustainable chemistry and pharmaceutical synthesis.

Related publication:

View Publications