Fig. 2
From: SVFX: a machine learning framework to quantify the pathogenicity of structural variants
Performance evaluation for somatic models to predict pathogenic SVs in various cancer types. This figure presents area auROCs based on the validation datasets for large deletions (a) and duplications (b) in six different cancer cohorts including breast adenocarcinoma (BRCA), esophageal carcinoma (ESCA), liver (LIHC), ovary (OV), skin melanoma (SKCM), and stomach (STAD) cancers. Similarly, auROC plots are shown for test datasets associated with large deletions (c) and duplications (d) in six different cancer cohorts. Finally, auROC plots are shown for pathogenic SVs in the validation (e) and testing dataset (f) in the ClinVar, CVD, and IBD cohort datasets