Fragment-Level Graph Encoding for Interpretable Molecular Property Prediction
Abstract
Graph neural networks (GNNs) are among the standard tools for molecular property prediction, but the atom-level explanations of their predictions are hard for medicinal chemists to act on. We propose a change of input representation rather than a new model: molecules are encoded as fragment-level graphs from BRICS decomposition, where each node is a synthetically meaningful fragment whose features are the element-wise sum of its atoms' features. Because only the representation changes, any message-passing GNN and any post-hoc attribution method produces explanations at the resolution of functional groups instead of single atoms. Across four architectures, four prediction tasks, and two explainers under scaffold splits, fragment-level encoding matches atom-level test accuracy on roughly five-fold smaller graphs. Integrated-Gradients attributions recover known pharmacophores above chance on two enzyme targets, namely BACE at 1.5 times the chance rate, and PTR1 at 3.2 times. A like-for-like comparison against an aggregated atom-level baseline shows that post-hoc aggregation is not a stable substitute: its BACE hit rate range both exceeds and falls below the fragment model's hit rate depending on two arbitrary aggregation choices, whereas fragment-level training yields a single convention-free attribution at no predictive cost. The code is available at: https://github.com/4D-Lab/frame
Related
- Numerical and Machine Learning Prediction of Blast and Fragment Propagation in RC Frames Under Internal Explosion
- What Comes Next and Why: Interpretable Next-Event Prediction with Neuro-Symbolic Rules
- BRP-GABLE: An Interpretable Rule-Based Prediction Framework for Multiple Outcome Types
- Interpretable Social Network Churn Prediction via Ergodic Markov Chains
- Development and external evaluation of an interpretable machine-learning model for early prediction of organ failure in higher-risk acute pancreatitis patients: A multicentre cohort study
BibTeX
@misc{almeida2026fragmentlevel,
title = {{Fragment-Level Graph Encoding for Interpretable Molecular Property Prediction}},
author = {Rafael Lopes Almeida and Ina Pöhner},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.29589},
eprint = {2610.29589},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.29589},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Fragment-Level Graph Encoding for Interpretable Molecular Property Prediction
AU - Rafael Lopes Almeida
AU - Ina Pöhner
PY - 2026
DA - 2026/10/21
PB - arXiv
DO - 10.48550/arxiv.2610.29589
UR - https://arxiv.org/abs/2610.29589
AB - Graph neural networks (GNNs) are among the standard tools for molecular property prediction, but the atom-level explanations of their predictions are hard for medicinal chemists to act on. We propose a change of input representation rather than a new model: molecules are encoded as fragment-level graphs from BRICS decomposition, where each node is a synthetically meaningful fragment whose features are the element-wise sum of its atoms' features. Because only the representation changes, any message-passing GNN and any post-hoc attribution method produces explanations at the resolution of functional groups instead of single atoms. Across four architectures, four prediction tasks, and two explainers under scaffold splits, fragment-level encoding matches atom-level test accuracy on roughly five-fold smaller graphs. Integrated-Gradients attributions recover known pharmacophores above chance on two enzyme targets, namely BACE at 1.5 times the chance rate, and PTR1 at 3.2 times. A like-for-like comparison against an aggregated atom-level baseline shows that post-hoc aggregation is not a stable substitute: its BACE hit rate range both exceeds and falls below the fragment model's hit rate depending on two arbitrary aggregation choices, whereas fragment-level training yields a single convention-free attribution at no predictive cost. The code is available at: https://github.com/4D-Lab/frame
ER -