Quantitative Toxicophore Discovery through Interpretable Deep Learning: A Framework for Mechanism-Guided Toxicity Mitigation in Drug Safety Profiling
- 1Yantai Institute of Science and Technology , , No.34 Xianjing West Road , Penglai District, , ,
Abstract
Unanticipated toxicity accounts for ∼30% of clinical-phase drug failures, yet despite advances in computational toxicity prediction, predictive models typically offer scores without identifying the structural culprit. Here, we introduce an interpretable deep-learning framework that couples fingerprint-descriptor fusion with task-adaptive attention and a SHAP-guided bioisosteric ablation pipeline. Trained on 9,315 compounds spanning 14 end points (Tox21 + ClinTox) under scaffold-stratified split, the attention-fusion model achieves mean test ROC-AUC of 0.739 (Tox21) and 0.953 (ClinTox). The critical advance is a structure-guided validation protocol: predictive intervention ablation─computationally replacing toxicophores with bioisosteric equivalents─demonstrates that the model’s predictions are predictively dependent on specific substructures. Halogen → methyl substitution drops predicted toxicity from 0.659 to 0.101 (paired t = 22.66, p = 6.32 × 10–10, n = 11; Cohen’s d = 6.82), providing the primary structure-guided evidence. Cross-end point SHAP correlation reaches 0.333 (MTL) and 0.334 (STL; Welch’s p = 0.82), indicating that both paradigms extract coherent mechanistic structure above the label baseline (0.172). SMARTS-based toxicophore alignment identifies a strong positive logit differential for aniline on the clinically relevant CT_TOX end point (+0.76), and predictive-intervention ablation confirms model dependence on the identified substructures (halogen → methyl: n = 11, p < 10–9; aniline → methoxy: n = 2, directional check only). External validation on 1,427 SIDER adverse-drug-reaction compounds reveals significant toxicophore enrichment (aniline: 12.6% vs 0%, Fisher exact p = 4.5 × 10–12; halogenated aromatic: 29.8% vs 0%, p = 1.6 × 10–29), remaining significant after controlling for molecular weight and Log P. Three case studies illustrate how SHAP-guided attribution translates into actionable medicinal-chemistry mitigation strategies.
Related
- DeepHeal: An Interpretable Framework for Phenotype-Based Antineoplastic Drug Discovery Using Drug-Perturbation Proteomics
- DISSOLVR: An Interpretable and Fast Framework for Aqueous and Organic Solubility Prediction
- An interpretable framework applying protein words to predict protein–small molecule complementary pairing rules
- Chemistry-infused bandit framework for interpretable autonomous campaign of electrocatalysis
- SandCat: OpenMM Framework for Prospective Cryptic Pocket Discovery
BibTeX
@article{gao2026quantitative,
title = {{Quantitative Toxicophore Discovery through Interpretable Deep Learning: A Framework for Mechanism-Guided Toxicity Mitigation in Drug Safety Profiling}},
author = {Jing Gao and Dong Lin and Jianfeng Hu},
journal = {ACS Omega},
year = {2026},
month = nov,
doi = {10.1021/acsomega.6c08546},
url = {https://pubs.acs.org/acsodf/article/doi/10.1021/acsomega.6c08546/5465435/Quantitative-Toxicophore-Discovery-through},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - Quantitative Toxicophore Discovery through Interpretable Deep Learning: A Framework for Mechanism-Guided Toxicity Mitigation in Drug Safety Profiling
AU - Jing Gao
AU - Dong Lin
AU - Jianfeng Hu
PY - 2026
DA - 2026/11/06
JO - ACS Omega
DO - 10.1021/acsomega.6c08546
UR - https://pubs.acs.org/acsodf/article/doi/10.1021/acsomega.6c08546/5465435/Quantitative-Toxicophore-Discovery-through
AB - Unanticipated toxicity accounts for ∼30% of clinical-phase drug failures, yet despite advances in computational toxicity prediction, predictive models typically offer scores without identifying the structural culprit. Here, we introduce an interpretable deep-learning framework that couples fingerprint-descriptor fusion with task-adaptive attention and a SHAP-guided bioisosteric ablation pipeline. Trained on 9,315 compounds spanning 14 end points (Tox21 + ClinTox) under scaffold-stratified split, the attention-fusion model achieves mean test ROC-AUC of 0.739 (Tox21) and 0.953 (ClinTox). The critical advance is a structure-guided validation protocol: predictive intervention ablation─computationally replacing toxicophores with bioisosteric equivalents─demonstrates that the model’s predictions are predictively dependent on specific substructures. Halogen → methyl substitution drops predicted toxicity from 0.659 to 0.101 (paired t = 22.66, p = 6.32 × 10–10, n = 11; Cohen’s d = 6.82), providing the primary structure-guided evidence. Cross-end point SHAP correlation reaches 0.333 (MTL) and 0.334 (STL; Welch’s p = 0.82), indicating that both paradigms extract coherent mechanistic structure above the label baseline (0.172). SMARTS-based toxicophore alignment identifies a strong positive logit differential for aniline on the clinically relevant CT_TOX end point (+0.76), and predictive-intervention ablation confirms model dependence on the identified substructures (halogen → methyl: n = 11, p < 10–9; aniline → methoxy: n = 2, directional check only). External validation on 1,427 SIDER adverse-drug-reaction compounds reveals significant toxicophore enrichment (aniline: 12.6% vs 0%, Fisher exact p = 4.5 × 10–12; halogenated aromatic: 29.8% vs 0%, p = 1.6 × 10–29), remaining significant after controlling for molecular weight and Log P. Three case studies illustrate how SHAP-guided attribution translates into actionable medicinal-chemistry mitigation strategies.
SN - 2470-1343
ER -