Fig. 3
From: Predicting variant deleteriousness in non-human species: applying the CADD approach in mouse
ROC-AUC scores of mCADD models evaluated on three different validation sets: the a) Fairfield, b) Mutagenetix and c) ClinVar-ESP data sets. The numbers below the bars indicate the number of annotations used during model training. Roman numbers indicate: I) the glire-PhastCons score, II) the vertebrate PhastCons score, and III) the hCADD* score. The numbers above the bars show the exact ROC-AUC of that particular model and validation set combination