Graph Energy-based Model for Substructure Preserving Molecular Design

TL;DR

This paper introduces GEM, a graph energy-based model that enables the generation of novel molecules while preserving specified substructures, integrating chemists' domain knowledge into molecular design.

Contribution

The paper presents a novel graph energy-based model that effectively preserves substructures during molecular generation, addressing limitations of existing methods.

Findings

GEM successfully generates molecules with preserved substructures.

The model produces novel molecules that align with chemists' domain knowledge.

Experimental results demonstrate the effectiveness of GEM on chemistry datasets.

Abstract

It is common practice for chemists to search chemical databases based on substructures of compounds for finding molecules with desired properties. The purpose of de novo molecular generation is to generate instead of search. Existing machine learning based molecular design methods have no or limited ability in generating novel molecules that preserves a target substructure. Our Graph Energy-based Model, or GEM, can fix substructures and generate the rest. The experimental results show that the GEMs trained from chemistry datasets successfully generate novel molecules while preserving the target substructures. This method would provide a new way of incorporating the domain knowledge of chemists in molecular design.