What Molecular Structure Cannot Tell Us: A Taxonomy of Explainability Gaps in GNN-Based Drug Toxicity Prediction

• URL: http://arxiv.org/abs/2605.26183v2
• Date: Fri, 29 May 2026 15:31:11 GMT
• Title: What Molecular Structure Cannot Tell Us: A Taxonomy of Explainability Gaps in GNN-Based Drug Toxicity Prediction
• Abstract summary: This study introduces an operational taxonomy of the structural information limits that prevent structure-based toxicity prediction.<n>A systematic case study using acetylsalicylic acid (ASA, Aspirin) serves as model compound.<n>Results indicate that molecular structure explains approximately 45% (5/11) of known ASA adverse effects.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Not all clinically relevant adverse effects are structurally inferable from molecular graphs - regardless of model quality or architectural complexity. This study introduces an operational taxonomy of the structural information limits that prevent structure-based toxicity prediction, independent of the learning algorithm employed. Graph Neural Networks (GNNs) have emerged as a natural approach for molecular toxicity prediction, operating directly on atomic connectivity without the information loss inherent to fixed-length fingerprints. However, the fraction of a drug's known pharmacological profile that is actually inferable from molecular structure remains systematically underexplored. A systematic case study using acetylsalicylic acid (ASA, Aspirin) - one of the most comprehensively characterized drugs in pharmacology - serves as model compound. A Message Passing Neural Network (MPNN) is trained on the Tox21 benchmark and GNNExplainer is applied to characterize atom-level attribution. Results indicate that molecular structure explains approximately 45% (5/11) of known ASA adverse effects. A four-category Gap Taxonomy (GAP-1 through GAP-4) is introduced distinguishing between principally non-encodable effects, data gaps arising from Missing Not At Random (MNAR) mechanisms, assay panel mismatches, and representation errors. The MNAR gap is empirically quantified via a systematic ChEMBL query (42 documented assays, 0 retrievable bioactivity entries). An attention pooling experiment localizes the representation error to the MPNN message passing layers rather than the aggregation step. The Gap Taxonomy has direct implications for drug safety signal detection and regulatory frameworks including Good Pharmacovigilance Practice (GVP) guidelines and New Approach Methodologies (NAMs). Structural limits identified are confirmed in a companion DDI ablation study.

Related papers

• RetiSEM: Generalising Causal Models for Fragmented Biomedical Data [18.644793391628234]
  RetiSEM is a framework for causal graph recovery and mediation analysis under limited multimodal resources.<n>We evaluate RetiSEM across ten synthetic benchmark scenarios that vary in dimensionality, nonlinearity, causal depth, and pathway structure.<n>In the real-data analysis, retinal variables behave mainly as downstream biomarker-like indicators, with smaller but detectable indirect effects.
  arXiv  Detail & Related papers  (2026-06-23T12:20:17Z)
• Pharmacogenomic Knowledge Graph Augmentation for Graph Neural Network-Based Drug-Drug Interaction Prediction [0.0]
  Graph neural networks (GNNs) applied to drug-drug interaction (DDI) prediction rely exclusively on molecular structure encoded as SMILES-derived graphs.<n>The present study investigates whether pharmacogenomic prior knowledge partially closes this gap.<n>Experiments are conducted under both pair-level and drug-level data splits to quantify generalization to unseen drugs.
  arXiv  Detail & Related papers  (2026-06-05T07:47:32Z)
• Learning Diagnostic Reasoning for Decision Support in Toxicology [34.623049546258294]
  We present DeToxR, the first adaptation of Reinforcement Learning (RL) to emergency toxicology.<n>We design a robust data-fusion engine for multi-label prediction across 14 substance classes.<n>We optimize the model's reasoning directly using a clinical performance reward.
  arXiv  Detail & Related papers  (2026-03-31T11:26:45Z)
• Empowering Chemical Structures with Biological Insights for Scalable Phenotypic Virtual Screening [13.925550352212412]
  This study presents DECODE, a framework that bridges the gap by empowering chemical representations with intrinsic biological semantics.<n>Our evaluations demonstrate that DECODE retrieves functionally similar drugs in zero-shot settings with over 20% relative improvement over chemical baselines in mechanism-of-action prediction.
  arXiv  Detail & Related papers  (2026-03-16T09:13:08Z)
• Aligned Manifold Property and Topology Point Clouds for Learning Molecular Properties [55.2480439325792]
  This work introduces AMPTCR, a molecular surface representation that combines local quantum-derived scalar fields and custom topological descriptors within an aligned point cloud format.<n>For molecular weight, results confirm that AMPTCR encodes physically meaningful data, with a validation R2 of 0.87.<n>In the bacterial inhibition task, AMPTCR enables both classification and direct regression of E. coli inhibition values.
  arXiv  Detail & Related papers  (2025-07-22T04:35:50Z)
• Regressor-free Molecule Generation to Support Drug Response Prediction [83.25894107956735]
  Conditional generation based on the target IC50 score can obtain a more effective sampling space. Regressor-free guidance combines a diffusion model's score estimation with a regression controller model's gradient based on number labels.
  arXiv  Detail & Related papers  (2024-05-23T13:22:17Z)
• Unveiling Molecular Moieties through Hierarchical Grad-CAM Graph Explainability [0.0]
  The integration of explainable methods to elucidate the specific contributions of molecular substructures to biological activity remains a significant challenge.<n>We trained 20 GNN models on a dataset of small molecules with the goal of predicting their activity on 20 distinct protein targets from the Kinase family.<n>We implemented the Hierarchical Grad-CAM graph Explainer framework, enabling an in-depth analysis of the molecular moieties driving protein-ligand binding stabilization.
  arXiv  Detail & Related papers  (2024-01-29T17:23:25Z)
• Learning to Denoise Biomedical Knowledge Graph for Robust Molecular Interaction Prediction [50.7901190642594]
  We propose BioKDN (Biomedical Knowledge Graph Denoising Network) for robust molecular interaction prediction. BioKDN refines the reliable structure of local subgraphs by denoising noisy links in a learnable manner. It maintains consistent and robust semantics by smoothing relations around the target interaction.
  arXiv  Detail & Related papers  (2023-12-09T07:08:00Z)
• Bi-level Contrastive Learning for Knowledge-Enhanced Molecule Representations [68.32093648671496]
  We introduce GODE, which accounts for the dual-level structure inherent in molecules.<n> Molecules possess an intrinsic graph structure and simultaneously function as nodes within a broader molecular knowledge graph.<n>By pre-training two GNNs on different graph structures, GODE effectively fuses molecular structures with their corresponding knowledge graph substructures.
  arXiv  Detail & Related papers  (2023-06-02T15:49:45Z)
• Graph neural networks for the prediction of molecular structure-property relationships [59.11160990637615]
  Graph neural networks (GNNs) are a novel machine learning method that directly work on the molecular graph. GNNs allow to learn properties in an end-to-end fashion, thereby avoiding the need for informative descriptors. We describe the fundamentals of GNNs and demonstrate the application of GNNs via two examples for molecular property prediction.
  arXiv  Detail & Related papers  (2022-07-25T11:30:44Z)
• Chemical-Reaction-Aware Molecule Representation Learning [88.79052749877334]
  We propose using chemical reactions to assist learning molecule representation. Our approach is proven effective to 1) keep the embedding space well-organized and 2) improve the generalization ability of molecule embeddings. Experimental results demonstrate that our method achieves state-of-the-art performance in a variety of downstream tasks.
  arXiv  Detail & Related papers  (2021-09-21T00:08:43Z)
• ASGN: An Active Semi-supervised Graph Neural Network for Molecular Property Prediction [61.33144688400446]
  We propose a novel framework called Active Semi-supervised Graph Neural Network (ASGN) by incorporating both labeled and unlabeled molecules. In the teacher model, we propose a novel semi-supervised learning method to learn general representation that jointly exploits information from molecular structure and molecular distribution. At last, we proposed a novel active learning strategy in terms of molecular diversities to select informative data during the whole framework learning.
  arXiv  Detail & Related papers  (2020-07-07T04:22:39Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.