Learning to SMILE(S)

TL;DR

The paper investigates applying deep learning directly to SMILES strings for ligand-based virtual screening in cheminformatics. It demonstrates that CNNs operating on SMILES can outperform traditional hand-crafted fingerprints, even on relatively small datasets, and can benefit from data augmentation via random SMILES walks. By drawing an analogy to NLP and sentiment analysis, the authors justify a text-based DL approach and show concrete performance gains over standard baselines. The work suggests a promising direction for leveraging large chemical databases with semi-supervised and unsupervised learning in cheminformatics.

Abstract

This paper shows how one can directly apply natural language processing (NLP) methods to classification problems in cheminformatics. Connection between these seemingly separate fields is shown by considering standard textual representation of compound, SMILES. The problem of activity prediction against a target protein is considered, which is a crucial part of computer aided drug design process. Conducted experiments show that this way one can not only outrank state of the art results of hand crafted representations but also gets direct structural insights into the way decisions are made.

Figures (3)

• Figure 1: SMILES produced for the compound in the figure is N(c1)ccc1N.
• Figure 2: Substituting highlighted carbon atom with nitrogen renders compound inactive.
• Figure 3: Visualization of CNN filters of size 5 for active (top row) and inactives molecules.