Finding low energy conformers is an important task in many aspects of computational chemistry. We previously developed a novel method, called the Bayesian Optimization Algorithm (BOA), to find the lowest energy conformation efficiently. In BOA, a probabilistic model is used to approximate the potential energy surface (PES). This allows us to quantify the model’s uncertainty and thus helps to guide sampling on the PES. This approach can avoid falling into local minima. However, this approach may sample conformers with steric clashes in the learning stage, so we developed a new acquisition function that incorporates knowledge of correlated torsions to avoid such problems. Bivariate von-Mises distributions were used to represent the torsion preferences of the correlated torsions, and we embedded this into our new acquisition function, which helps to focus on sampling in the low-energy regions. We tested our new approach, called BOKEI (Bayesian Optimization with Knowledge-based Expected Improvement) using two different energy functions: (i) a molecular mechanics force field, MMFF94; and, (ii) a semi-empirical method GFN2-xTB. We show that BOKEI searches significantly more efficiently than our original BOA method, requiring fewer energy evaluations to find the lowest energy conformer. More importantly, we show that correlated torsions not only describe steric clashes, but can also describe favorable intramolecular interactions, including intramolecular hydrogen bonds and π-π stacking. Further understanding of the conformational preferences of molecules will help accelerate the search for low energy conformers in a wide range of applications in computational chemistry.

BOKEI: Bayesian optimization using knowledge of correlated torsions and expected improvement for conformer generation

Macromolecular Chemistry: The Second Century

[Author Benefits](https://scimeetings.acs.org/?utm_source=pubs_content_marketing&utm_medium=website&utm_campaign=0320_MCF_NPI_Launch_Spring_Homepage&ref=pubs_content_marketing)

[How to use SciMeetings](https://storage.googleapis.com/dl.morressier.com/acs/SciMeetings-ACSSpring2020-HowToGuide_M3020-Final.pdf)

[Frequently Asked Questions](https://storage.googleapis.com/dl.morressier.com/acs/SciMtng_ACSNMSp2020_FAQ_M3120_Final.pdf)

ACS Spring 2020 National Meeting & Expo

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