Improving Predictions of Molecular Properties with Graph Featurisation and Heterogeneous Ensemble Models

TL;DR

This paper presents a hybrid modeling approach combining graph neural network features with traditional descriptors and ensemble methods, significantly improving molecular property predictions across various datasets.

Contribution

It introduces a MetaModel framework that effectively integrates diverse ML models and combines learned and conventional features for enhanced accuracy.

Findings

Outperforms ChemProp on all regression datasets

Outperforms ChemProp on 6 of 9 classification datasets

Including GNN features improves ensemble performance

Abstract

We explore a "best-of-both" approach to modelling molecular properties by combining learned molecular descriptors from a graph neural network (GNN) with general-purpose descriptors and a mixed ensemble of machine learning (ML) models. We introduce a MetaModel framework to aggregate predictions from a diverse set of leading ML models. We present a featurisation scheme for combining task-specific GNN-derived features with conventional molecular descriptors. We demonstrate that our framework outperforms the cutting-edge ChemProp model on all regression datasets tested and 6 of 9 classification datasets. We further show that including the GNN features derived from ChemProp boosts the ensemble model's performance on several datasets where it otherwise would have underperformed. We conclude that to achieve optimal performance across a wide set of problems, it is vital to combine general-purpose descriptors with task-specific learned features and use a diverse set of ML models to make the predictions.