{ "project": "hERG toxicity prediction - ExtraTrees-comprehensive DEPLOYMENT REFITS (third deployment model; Phase-5 best single model, refit on all labeled data)", "version": "deployment-refit-2026-11-16", "deployment_refit": true, "trained_on": "ALL labeled data: locked pools (cls 12,702 / reg 8,657) + masters remainder + absorbed external sets (hERG-70 462, new-compounds 967, FDA panel 36, REGFAIL 7, hERG-60, Test-1)", "training_files": { "classification": "cls_alldata.csv (14,441 rows)", "regression": "reg_alldata.csv (10,370 rows)" }, "hard_gate": "every external/regulatory compound verified present in the actual training files AND in the fitted training matrices by InChIKey connectivity (counts in validation_results.json; full report HARD_GATE_ET.md in the project record)", "models": { "cls_scaffold": { "file": "models/extratreesA_cls_morgan+maccs+desc+physchem_extra+herg+pharm_scaffold_DEPLOYMENT.pkl", "task": "classification", "protocol": "scaffold", "headline": true, "n_train": 14441, "n_features": 2499 }, "cls_random": { "file": "models/extratreesA_cls_morgan+maccs+desc+physchem_extra+herg+pharm_random_DEPLOYMENT.pkl", "task": "classification", "protocol": "random", "headline": false, "n_train": 14441, "n_features": 2499 }, "reg_scaffold": { "file": "models/extratreesA_reg_morgan+desc+herg+pharm_scaffold_DEPLOYMENT.pkl", "task": "regression", "protocol": "scaffold", "headline": true, "n_train": 10370, "n_features": 2331 }, "reg_random": { "file": "models/extratreesA_reg_morgan+desc+herg+pharm_random_DEPLOYMENT.pkl", "task": "regression", "protocol": "random", "headline": false, "n_train": 10370, "n_features": 2331 } }, "feature_blocks": { "classification": [ "morgan", "maccs", "desc", "physchem_extra", "herg", "pharm" ], "regression": [ "morgan", "desc", "herg", "pharm" ] }, "hyperparameters": { "n_estimators": 300, "n_jobs": 2, "random_state": 42, "note": "identical to the locked-pool Phase-5 ET finals (p5_approachA_et.py)" }, "protocol_note": "scaffold and random bundles wrap the same fitted model per task: identical combo/hyperparameters and all-data training make the protocol distinction vanish in deployment refits", "honest_locked_benchmark_scores": { "source": "Locked-pool Phase-5 refits (train+val only; every external/regulatory set locked out), results/phase5/PHASE5_RESULTS.md \u2014 these are the honest benchmark scores of this architecture. The DEPLOYMENT refits in this repo absorbed those sets and are never scored on them.", "classification_scaffold_refit": { "hERG-70": { "ACC": 0.76, "MCC": 0.543 }, "new_compounds_MCC": 0.149, "FDA_errors": "4/36", "REGFAIL_Tier1": "4/5" }, "classification_random_refit": { "hERG-70": { "ACC": 0.74, "MCC": 0.504 }, "new_compounds_MCC": 0.154, "FDA_errors": "5/36", "REGFAIL_Tier1": "4/5" }, "regression_scaffold_refit": { "hERG-70_RMSE": 0.838 }, "regression_random_refit": { "hERG-70_RMSE": 0.829 }, "note": "Best SINGLE (non-fusion) model on hERG-70 in the programme (MCC 0.543); the best overall classifier is Fusion G (0.548, XGBoost + CardioGenAI), which is not a deployment model." }, "sibling_deployment_repos": [ "Delower/herg-xgboost-deployment", "Delower/herg-cardiogenai-deployment" ], "warning": "DEPLOYMENT MODEL - trained on the regulatory/external sets. Do NOT report its scores on those sets as validation." }