Structural data for Protein-DNA complexes comprise ~5% of databases like the PDB. Considering their importance in gene regulation, a comprehensive understanding of protein-DNA interactions can have tremendous implications. Traditionally <i>in silico</i> methods relied on sequence-based approaches which fall short in structural contexts as many aspects of the interactions are dictated by the 3D shapes of the protein and DNA. DNA can often deform in a sequence dependent manner when bound by proteins. Even when including flexibility in docking studies, it is often insufficient to capture large deformations or capture sequence specificity. Herein we incorporate limited experimental data to guide molecular dynamics simulations in order to identify bound protein-DNA complexes at atomistic resolution. In doing so we use Bayesian Inference through a program called Modelling Employing Limited Data (MELD). MELD uses temperature and hamiltonian replica exchange to accelerate sampling and induce multiple binding and unbinding events while allowing conformational flexibility. We applied the method to study a small set of challenging protein-DNA systems extracted from a docking benchmark set, assessing success against the experimental structure. We are trying to optimize and generalize the methodology to be applicable to a wide variety of protein-DNA systems from different transcription factor families.

Prediction of DNA-protein complex structure using limited experimental data

ACS Spring Meeting 2021

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