Benchmarking machine-readable vectors of chemical reactions on computed activation barriers.

Autor: van Gerwen P; Laboratory for Computational Molecular Design, Institute of Chemical Sciences and Engineering, École Polytechnique Fédérale de Lausanne 1015 Lausanne Switzerland clemence.corminboeuf@epfl.ch.; National Center for Competence in Research-Catalysis (NCCR-Catalysis), École Polytechnique Fédérale de Lausanne 1015 Lausanne Switzerland., Briling KR; Laboratory for Computational Molecular Design, Institute of Chemical Sciences and Engineering, École Polytechnique Fédérale de Lausanne 1015 Lausanne Switzerland clemence.corminboeuf@epfl.ch., Calvino Alonso Y; Laboratory for Computational Molecular Design, Institute of Chemical Sciences and Engineering, École Polytechnique Fédérale de Lausanne 1015 Lausanne Switzerland clemence.corminboeuf@epfl.ch., Franke M; Laboratory for Computational Molecular Design, Institute of Chemical Sciences and Engineering, École Polytechnique Fédérale de Lausanne 1015 Lausanne Switzerland clemence.corminboeuf@epfl.ch., Corminboeuf C; Laboratory for Computational Molecular Design, Institute of Chemical Sciences and Engineering, École Polytechnique Fédérale de Lausanne 1015 Lausanne Switzerland clemence.corminboeuf@epfl.ch.; National Center for Competence in Research-Catalysis (NCCR-Catalysis), École Polytechnique Fédérale de Lausanne 1015 Lausanne Switzerland.
Jazyk: angličtina
Zdroj: Digital discovery [Digit Discov] 2024 Mar 07; Vol. 3 (5), pp. 932-943. Date of Electronic Publication: 2024 Mar 07 (Print Publication: 2024).
DOI: 10.1039/d3dd00175j
Abstrakt: In recent years, there has been a surge of interest in predicting computed activation barriers, to enable the acceleration of the automated exploration of reaction networks. Consequently, various predictive approaches have emerged, ranging from graph-based models to methods based on the three-dimensional structure of reactants and products. In tandem, many representations have been developed to predict experimental targets, which may hold promise for barrier prediction as well. Here, we bring together all of these efforts and benchmark various methods (Morgan fingerprints, the DRFP, the CGR representation-based Chemprop, SLATM d , B 2 R l 2 , EquiReact and language model BERT + RXNFP) for the prediction of computed activation barriers on three diverse datasets.
Competing Interests: There are no conflicts to declare.
(This journal is © The Royal Society of Chemistry.)
Databáze: MEDLINE
Externí odkaz: