Wednesday, 8th September, 2021, 13:00

Machine Learning Coarse Grained Models with Graph Neural Networks

Adrià Pérez will teach 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.

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.

Adria  Pérez 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.

Registration: https://register.gotowebinar.com/register/5159085993874550798

Speaker: Adrià Pérez, Computational Science Group - Research Programme on Biomedical Informatics (GRIB, IMIM/UPF)

Room Online