Machine learning for membrane simulations
The proposed research aims to apply the capabilities of machine learning to study molecular models of cell membranes. Cell membranes are composed of fluid lipid bilayers with a highly dynamic nature. By developing neural network models for the simulation of lipid bilayers, we seek to overcome the shortcomings of current computational methods, providing superior accuracy and efficiency. Coarse-grained neural network potentials (CG NNPs) are neural network models that have shown promising results in simulating macromolecules. The team led by Professor Cecilia Clementi at Freie Universität Berlin has made notable advancements in developing CG NNPs for proteins, showcasing their ability to predict potential energy surfaces accurately. This project focuses on extending this success to the domain of lipid bilayers, a field where our research group, led by Professor Peter Tieleman, has made significant contributions. Our goal is to create the first CG NNP for lipid bilayers, initially focusing on homogeneous membrane patches and subsequently expanding to representative lipid mixtures. The resulting publications and tools from this collaboration will showcase the two groups as pioneers in the field. In addition, modelling cell membranes with CG NNPs could streamline drug screening processes and potentially reduce the overall cost of developing new medications.
Voir la description complète du projetPeter Tieleman
Freie Universität Berlin
Computer science
Biotechnology; Artificial Intelligence; Pharmaceuticals
University of Calgary
Globalink Research Award