Beyond AlphaFold: The Next Wave of ML
Innovations in Protein Modelling
William McCorkindale
Abstract:
In the wake of the landmark achievements of AlphaFold, the
field of protein modelling is experiencing a proliferation of novel Machine
Learning (ML) techniques. This talk offers a comprehensive review of these
emerging methods, from the use of protein language models for structural
modelling (ESMFold) and function annotation (ProteInfer), to the application of
diffusion models for protein design (RFDiffusion) and protein-ligand docking
(DiffDock). I will also discuss anticipated developments on both the near-term
and distant horizons of this field, with examples of how these ML approaches
are being applied at CHARM Therapeutics to accelerate drug discovery.
Speaker bio:
William is an interdisciplinary scientist working on the
application of machine learning & cheminformatics for drug discovery in the
Modelling & Informatics team at CHARM. Funded by the Gates Scholarship,
William completed his PhD under Dr Alpha Lee at the University of Cambridge,
researching the application of deep learning for modelling chemical reactions,
bioactivity, and structure-based virtual screening. William previously worked
on extending the capabilities of AlphaFold as a Research Scientist intern at
Google DeepMind.