🧬 MERGE Pathogenicity Prediction
Scores from all sub-models are assembled as features (multi-transcript values are averaged), missing values are imputed using MICE based on training distribution, then MinMax normalized. The final pathogenicity probability is computed by the MERGE model.
Running MERGE model & computing SHAP values…
📊 Extracted Feature Values
(yellow border = missing → imputed)
Evo2 (Local) Results
Evo2 queries the local model with the upstream genomic context to obtain scores. LLR = Reference Score - Variant Score.