Linear Graphlet Models for Accurate and Interpretable Cheminformatics

Advances in machine learning have given rise to a plurality of data-driven methods for estimating chemical properties from molecular structure. For many decades, the cheminformatics field has relied heavily on structural fingerprinting, while in recent years much focus has shifted leveraging highly parameterized deep neural networks which usually maximize accuracy. Beyond accuracy, machine learning techniques need intuitive and useful explanations for the predictions of models and uncertainty quantification techniques so that a practitioner might know when a model is appropriate to apply to new data. Here we show that linear models built on unfolded molecular-graphlet-based fingerprints attain accuracy that is competitive with the state of the art while retaining an explainability advantage over black-box approaches. We show how to produce precise explanations of predictions by exploiting the relationships between molecular graphlets and show that these explanations are consistent with chemical intuition, experimental measurements, and theoretical calculations. Finally we show how to use the presence of unseen fragments in new molecules to adjust predictions and quantify uncertainty.