Multimodal Cross-Attentive Graph-Based Framework for Predicting In Vivo Endocrine Disruptors
Santos, E. S. d. A.; Felizardo, G. S. S.; da Silva, A. C. G.; Martin, H.-J.; Muratov, E. N.; Braga, R. d. C.; Neves, B. J.
Show abstract
Endocrine hazard assessment needs models that are accurate and mechanistically transparent. We present a multimodal cross-attentive graph framework that fuses molecular graphs with adverse-outcome-pathway (AOP)-anchored assay signals to predict organism-level outcomes in the OECD Hershberger and uterotrophic assays. In Tier-1, multitask GNNs learn ER/AR molecular-initiating and key events across 46 ToxCast/Tox21 assays. In Tier-2, a cross-attentive multimodal GNN integrates Tier-1 pathway signals with molecular graphs, achieving AUROC{square}={square}0.90 (Hershberger) and 0.96 (uterotrophic). External validation on literature compounds showed 84% concordance (Hershberger 15/18; uterotrophic 22/26). Bidirectional cross-attention links molecular substructures to pathway assays and vice-versa, while counterfactual perturbations rank assays and structural motifs most responsible for each decision. The framework couples high accuracy with assay-traceable explanations, supporting targeted testing within the Integrated Approaches.
Matching journals
The top 6 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Beyond the nucleus: Plastic chemicals activate G protein-coupled receptors 93%
- Acidity of expiratory aerosols controls the infectivity of airborne influenza virus and SARS-CoV-2 89%
- Characterization of annual average traffic-related air pollution levels (particle number, black carbon, nitrogen dioxide, PM 2.5 , carbon dioxide) in the greater Seattle area from a year-long mobile monitoring campaign 89%
Similar papers in this journal
- Adding stochastic negative examples into machine learning improves molecular bioactivity prediction 95%
- Compound activity prediction with dose-dependent transcriptomic profiles and deep learning 95%
- BOLD-GPCRs: A Transformer-Powered App for Predicting Ligand Bioactivity and Mutational Effects Across Class A GPCRs 95%
Similar papers in this journal
- Multi-behavioral phenotyping in early-life-stage zebrafish for identifying disruptors of non-associative learning 94%
- Genetic variability in pathways associates with pesticide-induced nervous system disease in the United States 90%
- A Data-Driven Transcriptional Taxonomy of Adipogenic Chemicals to Identify White and Brite Adipogens 87%
Similar papers in this journal
- A Method To Calibrate Chemical Agnostic Quantitative Adverse Outcome Pathways On Multiple Chemical Dose-Response Data 91%
- Systematic analysis of protein targets associated with adverse events of drugs from clinical trials and post-marketing reports 89%
- Cell Painting and chemical structure read-across can complement each other for rat acute oral toxicity prediction in chemical early de-risking 88%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.