Molecule Generation from Input-Attributions over Graph Convolutional Networks

TL;DR

This paper introduces a method combining Graph Convolutional Networks and input-attribution techniques to generate new molecules, addressing challenges like over-optimization and applicability in drug design.

Contribution

It presents an automatic molecule generation process using GCNs and input-attribution, advancing tools for drug discovery and molecular optimization.

Findings

Effective molecule generation from input attributions

Identification of over-optimization issues

Insights into applicability limitations

Abstract

It is well known that Drug Design is often a costly process both in terms of time and economic effort. While good Quantitative Structure-Activity Relationship models (QSAR) can help predicting molecular properties without the need to synthesize them, it is still required to come up with new molecules to be tested. This is mostly done in lack of tools to determine which modifications are more promising or which aspects of a molecule are more influential for the final activity/property. Here we present an automatic process which involves Graph Convolutional Network models and input-attribution methods to generate new molecules. We also explore the problems of over-optimization and applicability, recognizing them as two important aspects in the practical use of such automatic tools.