Skip to main content
. 2020 Apr 13;36(12):3637–3644. doi: 10.1093/bioinformatics/btaa242

Fig. 7.

Fig. 7.

Performance of the best CScape-somatic model with the original CScape, CADD and TransFIC models on the ICGC test set for coding regions. Top: CScape-somatic yields substantially higher accuracy, at over 63%, than the original CScape at 59%. The remaining methods fare less well, with accuracies ranging from 51 to 56%. Bottom: When consider ranking performance, CScape-somatic again outperforms the other methods with an AUC score of 0.69, followed by the original CScape at 0.61 and the remaining methods all below 0.60. (CSS= CScape-somatic; CS= CScape; TF-MAS= TransFIC-MutationAssessor, TF-PPH2= Transfic-Polyphen2 and TF-SIFT= TransFIC-SIFT.)