Difficulty in chirality recognition for Transformer architectures learning chemical structures from string
This work addresses a black-box issue in applying NLP models to chemistry, providing insights into learning bottlenecks for researchers in cheminformatics and AI, though it is incremental in nature.
The study tackled the problem of how Transformer architectures learn chemical structures from SMILES strings, revealing that while partial structures are learned quickly, overall structures and chirality recognition require extended training, with chirality sometimes causing performance stagnation due to enantiomer misunderstandings.
Recent years have seen rapid development of descriptor generation based on representation learning of extremely diverse molecules, especially those that apply natural language processing (NLP) models to SMILES, a literal representation of molecular structure. However, little research has been done on how these models understand chemical structure. To address this black box, we investigated the relationship between the learning progress of SMILES and chemical structure using a representative NLP model, the Transformer. We show that while the Transformer learns partial structures of molecules quickly, it requires extended training to understand overall structures. Consistently, the accuracy of molecular property predictions using descriptors generated from models at different learning steps was similar from the beginning to the end of training. Furthermore, we found that the Transformer requires particularly long training to learn chirality and sometimes stagnates with low performance due to misunderstanding of enantiomers. These findings are expected to deepen the understanding of NLP models in chemistry.