Physical Society Colloquium

Learning from Life: Understanding and Design of Complex Biophysical Systems through Multiscale Modeling and Machine Learning

Ré Mansbach

Department of Physics
Concordia University

What do hybrid antibiotics and venomous snail proteins have in common? Both are complex biological systems ripe for computational physics design. In this talk, I will explain my research interests in physics-based design of complex biophysical problems and briefly situate them in the context of three recent projects. First, I will show how developing a fragment-based machine learning algorithm in conjunction with coarse-grained molecular dynamics ("Hunting FOX" for "Hunting Fragments Of X") led to understanding the physical mechanisms by which different chemical submolecules contribute to antibiotic permeation into Gram-negative bacteria. This represents a step towards hybrid antibiotic design to combat antimicrobial drug resistance. Second, I will demonstrate how I am using nonlinear dimensionality reduction and atomistic molecular dynamics simulations to map out the folding landscape and eventually design space of disulfide-rich peptides. Finally, I will briefly discuss the insights molecular dynamics and a simple Ising model have given for understanding of the SARS-CoV-2 Spike trimer. Overall, I will give a broad overview of my interests and where I foresee those interests taking me and my lab.