The Free Energy Landscape (FEL), a representation of the energetic configurational space underlying the thermodynamics and kinetics of any molecular process, can be used to describe the conformational substate selection processes of molecules. Therefore in the case of a mutated molecule, if the FEL can be restored to an active state, a major challenge to drug design, our working hypothesis is that the activity of the protein will also be restored. The recent work in our lab has been aimed at building tools and gaining foundational knowledge pertaining to the conformational dynamics of proteins, particularly p53. This project focuses on capturing the energetic interactions amongst p53 residues to be used for machine learning to restore native functionality to mutated proteins. The FEL can be captured and visualized using three dimensional energy vectors crafted with data from Molecular Dynamics (MD) simulations in order to represent the net pairwise interaction energy. Each of these energy vectors is then overlaid onto the alpha-carbon of their respective residues to create a visual backbone and representation of the FEL. Using assisted machine learning and energetic training data, these vectors can be used as inputs in a linear regression model for predicting the conformational substate selection of a molecule.

Machine learning on the free energy landscape of p53 captured with energy vectors from MD simulations

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.