Research Projects

Molecular dynamics (MD) has revolutionized the landscape of chemistry, biology and material science by elucidating atomistic insights into molecular interactions in different relevant processes. However requirements of small integration time steps of standard brute-force MD, biologically relevant processes like protein folding transition, large conformational transition, ligand binding-unbinding and processes with high energy barrier that occur on microsecond to millisecond timescales remain challenging to capture through standard unbiased MD. To address these limitations, variety of enhanced sampling methods have been developed over the last two decades where methods based on collective variables (CVs) have proven particularly effective in biasing the system along relevant reaction coordinates (RC), thereby promoting transitions and improving sampling of rare events. However, selecting appropriate CVs for bio-molecular processes remains challenging and this problem is yet to be solved comprehensively. Very recently, we have developed a method to estimate the relevance of different CVs near transition-state ensembles (TSEs). Building on this foundation, in this project we will first formulate a proper CV from TSEs through our proposed explainable artificial intelligence (AI) framework and build a novel enhanced sampling method using that TSE-inspired CV. We will investigate mechanism of polymer collapse transition for higher chain lengths, protein folding including relatively bigger protein, structural transitions in metamorphic proteins, drug unbinding and conformational transition coupled with ligand binding and inhibition of protein aggregation by small molecules using our proposed explainable AI based enhanced sampling method integrated with Markov state model (MSM) and generative AI framework. The generative AI component will be employed to generate new physically plausible conformations in low data regimes thereby improving sampling efficiency and capturing transient states critical to bio-molecular mechanism. An effort will be made to develop general CVs for a type of biological process that can be transferrable across systems. Deciphering the mechanism of protein-folding transition and protein-drug bindingis of immense interest for experimental biologists as well as biophysics community. Moreover, protein aggregation is responsible for several neurodegenerative diseases underscoring its biomedical relevance of this proposal. Successful completion of this project provides effective CVs to study these complex bio-molecular processes, offer valuable insights into the mechanism of complex processes and kinetics involved between different crucial intermediates.