PathGennie- an open source computational framework from researchers at S. N. Bose National Centre for Basic Sciences promises to democratise high-accuracy drug unbinding simulations.
PathGennie, a newly developed open source computational framework, is set to significantly accelerate computer-aided drug discovery by making high-fidelity drug–protein unbinding simulations broadly accessible to the global scientific community.
Freely available and openly extensible, PathGennie addresses one of the most persistent challenges in molecular modelling: simulating rare molecular events, particularly the unbinding of drug molecules from protein targets. This process is central to predicting drug residence time, a parameter increasingly recognised as more critical to therapeutic efficacy than binding affinity alone.
Conventional molecular dynamics simulations struggle with such events because they occur over long timescales. Existing approaches often rely on artificial bias forces or elevated temperatures to speed up simulations, but these interventions can distort the underlying physics. PathGennie eliminates this limitation by preserving true kinetic pathways without applying external biases.
The framework is built on a Direction-Guided Adaptive Sampling algorithm inspired by natural selection. It launches large swarms of ultrashort, unbiased molecular dynamics trajectories and selectively extends only those that move closer to a desired end state, discarding unproductive paths. This enables rare-event pathways to be identified efficiently without long waiting times.
In proof-of-concept studies, PathGennie successfully uncovered multiple competing unbinding routes, including benzene escaping from the T4 lysozyme binding pocket and three distinct dissociation pathways of the anti-cancer drug imatinib from Abl kinase. All previously reported pathways were recovered within a few iterations, matching earlier simulations and experimental results.
Developed at the S. N. Bose National Centre for Basic Sciences, Kolkata, under the Department of Science and Technology (DST), the work was led by Prof. Suman Chakrabarty, with Dibyendu Maity and Shaheerah Shahid. The research is published in the Journal of Chemical Theory and Computation, positioning India as a contributor to foundational open scientific software.
