Enhancing Chemical Explainability Through Counterfactual Masking

TL;DR

This paper introduces counterfactual masking, a new method for explaining molecular property predictions by replacing masked substructures with realistic fragments, leading to more meaningful and distribution-consistent explanations.

Contribution

The paper proposes a novel counterfactual masking framework that improves molecular explainability by ensuring realistic substitutions using generative models, unlike traditional masking methods.

Findings

Counterfactual masking produces more realistic explanations.

The method enhances benchmarking of explainers across datasets.

It offers actionable insights for molecular design.

Abstract

Molecular property prediction is a crucial task that guides the design of new compounds, including drugs and materials. While explainable artificial intelligence methods aim to scrutinize model predictions by identifying influential molecular substructures, many existing approaches rely on masking strategies that remove either atoms or atom-level features to assess importance via fidelity metrics. These methods, however, often fail to adhere to the underlying molecular distribution and thus yield unintuitive explanations. In this work, we propose counterfactual masking, a novel framework that replaces masked substructures with chemically reasonable fragments sampled from generative models trained to complete molecular graphs. Rather than evaluating masked predictions against implausible zeroed-out baselines, we assess them relative to counterfactual molecules drawn from the data distribution. Our method offers two key benefits: (1) molecular realism underpinning robust and distribution-consistent explanations, and (2) meaningful counterfactuals that directly indicate how structural modifications may affect predicted properties. We demonstrate that counterfactual masking is well-suited for benchmarking model explainers and yields more actionable insights across multiple datasets and property prediction tasks. Our approach bridges the gap between explainability and molecular design, offering a principled and generative path toward explainable machine learning in chemistry.