Fig. 5: MAVERICK reliably prioritizes causal variants in real cases.

a–c Waffle plots showing rank of causal variant in 644 real cases solved by MAVERICK’s rank ordering of variants with inheritance and phenotypic information incorporated. a MAVERICK performance on 130 cases whose causal variant was in MAVERICK’s training set. Cases where either variant in a compound heterozygous pair are in the training set are included here. b MAVERICK performance on 375 cases whose causal variant was not in the training set, but at least one pathogenic variant on that gene was in the training set. c MAVERICK performance on 139 cases whose causal variant was on a gene without any pathogenic variants in the training set. d Cumulative proportion of real cases solved by MAVERICK’s rank ordering of variants for 36 cases whose causal variants were on disease genes that were recently discovered and novel to MAVERICK. The CADM3 and PRDX3 cases are all solved on the first guess and so the CADM3 line is obscured by the PRDX3 line. e Stacked horizontal bar plots of the cumulative proportion of 18 patients with causal variants on novel disease genes solved by MAVERICK or Exomiser’s rank ordering of variants. Results are shown using genotype information only as well as with phenotypic information incorporated.