Fig. 2

Performance of tested methods on region-matched pathogenic and non-pathogenic ncSNVs. a Violin plots show distributions of impact score difference between nearby pathogenic and non-pathogenic ncSNVs. Variants were grouped into bins based on distances measured in base pairs. Positive values imply that the pathogenic variant has the higher score. b Correlation of impact scores for ncSNVs located within given genomic distances. Pearson correlation coefficient values are displayed. c Success rates of ranking a pathogenic ncSNV higher than a non-pathogenic ncSNV located within 1000 bps vicinity. Data are stratified by conservation of the position harboring pathogenic ncSNVs. Since pathogenic and non-pathogenic variants are evaluated in pairs, the random expectation of success rates is 50% as represented by the red line. d Success rates stratified by the genomic context of pathogenic ncSNVs, including promoters, 5’- and 3’- untranslated regions (UTRs), introns, near-gene (+/−100 kbps) regions and gene-desert regions. GWAVA represents GWAVA region-matched scores