CompBioMed’s 17th e-seminar took place on 8 September 2021
Machine Learning Coarse-Grained Models with Graph Neural Networks.
Atomistic or ab initio molecular dynamics simulations are widely used to predict thermodynamics and kinetics and relate them to molecular structure. A common approach to go beyond the time-and length-scales accessible with such computationally expensive simulations is the definition of coarse-grained molecular models. Recent advances in machine learning allow us to design optimal coarse-grained models for reproducing the equilibrium thermodynamics of a macromolecule.
In this seminar, we will learn how to prepare, train and simulate coarse-grained models with graph neural networks using TorchMD, a deep learning framework for molecular simulations. We will build a coarse-grained model of the fast-folding protein chignolin in order to simulate it and reproduce its folding process.
The e-seminar was given by Adrià Pérez (GRIB, UPF).
Adria is a Research Fellow at the Computational Science group, GRIB, UPF. He holds a Biochemistry degree from Universitat Pompeu Fabra (UPF). The computational science research group, led by Gianni de Fabritiis, is dedicated to computational science in biomedicine and machine learning. The group research interested are rooted in application of computation to solve real world problems. Specifically, they develop new methods and algorithms and apply them to computational chemistry, drug design, protein folding, etc. The group and the spin-off company Acellera, founded in 2006, has collaborated with major industries worldwide like Sony NVIDIA, HTC mobile, UCB and Pfizer.
This webinar series is run in collaboration with the VPH Institute
