CandidateDrug4Cancer: An Open Molecular Graph Learning Benchmark on Drug Discovery for Cancer

TL;DR

CandidateDrug4Cancer is a comprehensive benchmark dataset for graph machine learning aimed at improving anti-cancer drug discovery, highlighting challenges and opportunities in predicting drug-target interactions.

Contribution

The paper introduces a large, realistic benchmark dataset for molecular graph learning in cancer drug discovery and evaluates baseline methods on this dataset.

Findings

The dataset covers 29 cancer targets and 54,869 drug molecules.

Baseline models show significant challenges in predicting drug-target interactions.

Results indicate room for improvement in graph neural network performance for this application.

Abstract

Anti-cancer drug discoveries have been serendipitous, we sought to present the Open Molecular Graph Learning Benchmark, named CandidateDrug4Cancer, a challenging and realistic benchmark dataset to facilitate scalable, robust, and reproducible graph machine learning research for anti-cancer drug discovery. CandidateDrug4Cancer dataset encompasses multiple most-mentioned 29 targets for cancer, covering 54869 cancer-related drug molecules which are ranged from pre-clinical, clinical and FDA-approved. Besides building the datasets, we also perform benchmark experiments with effective Drug Target Interaction (DTI) prediction baselines using descriptors and expressive graph neural networks. Experimental results suggest that CandidateDrug4Cancer presents significant challenges for learning molecular graphs and targets in practical application, indicating opportunities for future researches on developing candidate drugs for treating cancers.