<br/>Automation is essential in modern computational research, and can open up possibilities unimaginable under the manual approach. We combine two computational directions of research related to automation to create a series of unique algorithms for reactivity studies. One is a machine learning-based optical recognition of chemical structures that allows one to build three-dimensional models of molecules from recent chemical publications, and thus of particular interest to chemical community. The other one is an automated transition state search which is designed to find a transition state between two user-specified potential energy minima, and thus predict kinetics and thermodynamics of the chemical transformation. Combined, they create a tool for an automated stress-testing and honing transition state search algorithms and building up libraries of transition states relevant to modern chemical research in such areas as catalyst and drug design.

Machine learning-based structure recognition for automated reactivity predictions through transition state calculations

255th National Meeting (2018)

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