WEHI New Medicines and Diagnostics Special Seminar hosted by Dr Rohan Volpe

Dr Thomas Coudrat

Computational Chemist and Research Scientist – Protein and Molecular Modelling Team, CSIRO

Computational drug design at CSIRO

Davis Auditorium | Bundoora Small Boardroom
Join via TEAMS
Including Q&A session

Computational chemistry aims to support the early-stage drug discovery phases of hit identification, hit to lead and lead optimisation. I will describe our approach to identify novel hits as well as current developments in hit to lead and lead optimisation. Classical virtual screening (VS) uses physics-based computer modelling to identify small molecule hits by searching molecule libraries. In parallel, ultra-large virtual libraries like the Enamine REAL database (6.75 billion molecules) are providing new opportunities to identify novel chemical matter. Classical VS by docking isn’t achievable on such scale, therefore we’ve implemented an active learning approach where we iteratively dock subsets of the virtual library to train a machine learning (ML) model that becomes a surrogate of the docking function. This new surrogate ML model is orders of magnitude faster that the docking software which enables VS on the scale of the Enamine REAL database. I’ll also present two exciting developments in hit to lead and lead optimisation. The first is the application of free energy perturbation (FEP) to predict the potency of novel molecules within a series to guide structure-activity (SAR) development. And the second is the deployment of ML models to predict absorption, distribution, metabolism, and excretion (ADME) properties of drug-like molecules.

Thomas is an industry-focused computational chemist. He completed his PhD in computational chemistry at MIPS in Parkville working on virtual screening of G protein-coupled receptors. He worked for 2.5 years at Galapagos in Paris, a mid-sized biotech company where he supported medicinal chemistry projects with a range of computational chemistry techniques. During this time, he also developed and put in production a machine learning model for ADME property prediction. Upon his return to Australia, he headed computational chemistry for 2 years in a spinout company from Monash University. He has now been working at CSIRO for over 2 years, where he provides computational chemistry support to clients and leads the internal development of computational chemistry and AI for drug design capabilities. 

All welcome!