Talk: "Deep learning-based prediction of inter-residue contacts and tertiary structure in proteins"
Venue: MIB Lecture Theatre, The University of Manchester
Sorry, this event has now ended.
Speaker:Dr Shaun Kandathil (the Francis Crick Institute)
Title: "Deep learning-based prediction of inter-residue contacts and tertiary structure in proteins"
Analyses of amino-acid residue covariation in large multiple sequence alignments have led to breakthroughs in template-free or de novoprotein structure prediction. More recently, deep learning techniques have been applied to this area, further advancing our ability to model protein structures de novo. In this talk, I will describe our work applying deep learning in two areas of protein structure prediction, namely inter-residue contact prediction and complete de novo modelling of protein structures. In the latter case, our method DMPfold is capable of fast and accurate modelling of protein structures on a genomic scale, using relatively modest computing resources. Applying the method to a set of protein families without template structures available, we find that we are able to extend structural coverage for many proteomes, far beyond what was previously possible. All the methods I will describe are freely available.