PTP1B (protein tyrosine phosphatase 1B) is a negative regulator of the insulin pathway and its inhibition would have implications for diabetes and obesity treatment. However, a small molecule inhibitor specific enough for pharmaceutical use on this charged family of proteins has been elusive. To determine<i> </i>computationally<i> </i>a potential allosteric inhibitor of PTP1B, location coordinates from high-throughput docking to various molecular dynamics-generated conformations were collected and processed through a random forest machine learning model. The random forest model suggested most docking variability occurs at the alpha-7 helix site, supporting existing literature. The docking results of positive controls were compared to their enantiomers, showing distributions of docking scores that were highly similar and demonstrated further the need for methodological development. Statistical comparisons of docking score distributions as well as machine learning analysis results are currently being developed as inputs for the custom scoring function of the genetic algorithm application, AutoGrow 4. This can then be used to construct and rank synthesizable inhibitors to efficiently guide efforts currently underway to synthesize and empirically test allosteric inhibitors of PTP1B.

Application of machine learning and genetic algorithms for the computational design of allosteric inhibitors of PTP1B

Where thousands of chemistry professionals meet to share ideas and advance scientific and technical knowledge. Spring 2023 abstracts, posters, and presentations will launch at the end of March.

ACS Spring 2023

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