Fig. 3From: Introme accurately predicts the impact of coding and noncoding variants on gene splicing, with clinical applicationsIntrome performance comparison. A Breakdown of the location of splice-altering variants making up the validation set. The distance of the regions to respective intron–exon boundaries is in brackets. B Precision recall curve of several splice predictor programs on the validation set. The area under the curve (auPRC) for each tool is shown in the legend. The dotted line at 0.95 precision indicates the threshold used to evaluate performance in C. C Performance recall for each evaluated program across the multiple datasets by variant region. Thresholds used represent 0.95 precision on the validation datasetBack to article page