University of Massachusetts Medical School Faculty Publications
UMMS Affiliation
Department of Biochemistry and Molecular Pharmacology; Schiffer Lab
Publication Date
2020-06-09
Document Type
Article Preprint
Disciplines
Artificial Intelligence and Robotics | Biophysics | Enzymes and Coenzymes | Medicinal Chemistry and Pharmaceutics | Molecular Biology | Structural Biology
Abstract
Drug resistance threatens many critical therapeutics through mutations in the drug target. The molecular mechanisms by which combinations of mutations, especially involving those distal from the active site, alter drug binding to confer resistance are poorly understood and thus difficult to counteract. A strategy coupling parallel molecular dynamics simulations and machine learning to identify conserved mechanisms underlying resistance was developed. A series of 28 HIV-1 protease variants with up to 24 varied substitutions were used as a rigorous model of this strategy. Many of the mutations were distal from the active site and the potency to darunavir varied from low pM to near μM. With features extracted from molecular dynamics simulations, elastic network machine learning was applied to correlate physical interactions at the molecular level with potency loss. This fit to within 1 kcal/mol of experimental potency for both the training and test sets, outperforming MM/GBSA calculations. Feature reduction resulted in a model with 4 specific features that correspond to interactions critical for potency regardless of enzyme variant. These predictive features throughout the enzyme would not have been identified without dynamics and machine learning and specifically varied with potency. This approach enables capturing the conserved dynamic molecular mechanisms by which complex combinations of mutations confer resistance and identifying critical interactions which serve as bellwethers over a wide range of inhibitor potency. Machine learning models leveraging molecular dynamics can thus elucidate mechanisms that confer loss of affinity due to variations distal from the active site, such as in drug resistance.
Keywords
Biophysics, drug resistance, molecular dynamics, machine learning
Rights and Permissions
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
DOI of Published Version
10.1101/2020.06.08.139105
Source
bioRxiv 2020.06.08.139105; doi: https://doi.org/10.1101/2020.06.08.139105. Link to preprint on bioRxiv service.
Journal/Book/Conference Title
bioRxiv
Repository Citation
Leidner F, Yilmaz NK, Schiffer CA. (2020). Deciphering complex mechanisms of resistance and loss of potency through coupled molecular dynamics and machine learning [preprint]. University of Massachusetts Medical School Faculty Publications. https://doi.org/10.1101/2020.06.08.139105. Retrieved from https://escholarship.umassmed.edu/faculty_pubs/1705
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Included in
Artificial Intelligence and Robotics Commons, Biophysics Commons, Enzymes and Coenzymes Commons, Medicinal Chemistry and Pharmaceutics Commons, Molecular Biology Commons, Structural Biology Commons