Bidirectional Graphormer for Reactivity Understanding: neural network trained to reaction atom-to-atom mapping task

Abstract

This work introduces GraphormerMapper – a new algorithm for reactions atom-to-atom mapping (AAM) based on a distance-aware BERT neural network. In benchmarking studies with IBM RxnMapper, the best AAM algorithm according to our previous study, we demonstrate that our AAM algorithm is superior on our “Golden” benchmarking dataset. The mapper is implemented in Chython [https://github.com/chython/chython] and Chytorch [https://github.com/chython/chytorch, https://github.com/chython/chytorch-rxnmap] Python packages which are freely available for out-the-box use. Chython is a cheminformatics library with a simple interface for processing reaction and molecular data. The key features of Chython are: chemical functional groups standardization, checking atom valence errors, substructure search, and advanced reaction manipulation, for example, generating products from reactants and reaction atom-to-atom mapping. Chytorch provides a PyTorch-like interface for graph-based neural networks developed specifically for chemical tasks.

Comparison of Atom Maps

“…This result provides a rigorous justification for the computational approach in [28] to compare atom maps by testing for isomorphisms of their ITSs (or CGRs). Nevertheless we should add that the authors of [28] and [36] introduced and made use of a manually curated set of reactions for their benchmarking purposes, which they refer to as Golden data set and which consists of both stoichiometrically balanced and unbalanced reactions. In our contribution we also made use of this data set for our own computational analysis, but we only employed the former type of reactions, since it should be noted that the use of the latter type is in conflict with the formal requirement now established by Cor.…”

“…GraphormerMapper [36] is also based on a Transformer Neural Network architecture. The EEquAAM suite also provides a wrapper for these tools to interact with the comparison methods mentioned above.…”

“…The EEquAAM suite also provides a wrapper for these tools to interact with the comparison methods mentioned above. To assess the performance of the different methods implemented in EEquAAM we first used a subset of the Golden data set, which was originally collected with the aim of benchmarking atom mapping tools [28,36].…”