Limitations of representation learning in small molecule property prediction

Abstract

Representation learning is making inroads into drug discovery. A study in Nature Communications emphasizes multiple limitations in property prediction. The results suggest that continued research and improvements are required for this specific area that coalesces machine learning and molecular medicine. Machine learning is a powerful tool for the study and design of molecules. Here the authors comment a recent publication in Nature Communications which highlights the challenges of different molecular representations for data-driven property predictions.