Extended Data Fig. 7: Investigating features of events seen/not seen in RNA studies for our 88 variants.

True positive (TP) events seen in RNA studies (n = 119), versus false positives (FP) in the Top-4 (n = 238) have higher: a) sample counts (W = 20,222, p = 4e⁻¹¹), b) max (W = 20,144, p = 7e⁻¹¹), and c) mean reads (W = 18,284, p = 7e^-6, two-sided Wilcoxon rank sum test). Events are biologically independent in A-C. For the annotated splice-junction around which the unannotated event is detected, TP, versus FP have d) no significant difference in the mean read-depth (W = 12,848, p = 0.3), e) higher maximum ratio (W = 19,126, p = 1e⁻⁹) but f) no significant difference in the mean ratio of unannotated to annotated reads (W = 15,181, p = 0.1, two-sided Wilcoxon rank sum test). n = 234 Top-4* FP and 117 TP biologically independent events for (D-F): 4/238 FP and 2/119 TP were detected only in samples where no annotated splicing was detected, so are excluded from d-f. g) Single-exon skipping events* are significantly more likely to be seen in RNA studies than double-exon skipping events* (Chi-squared test; 𝝌² = 114.29, p = 1.1e⁻²⁶). h) Total length (nt) of the fragment excised from the pre-mRNA from single and double exon skipping was not statistically different between TP and FP (two-sided Wilcoxon rank sum test; W = 2,554, p = 0.0502, n = 114 Top-4* FP and 75 TP biologically independent exon-skipping events). i) Distance from annotated splice-site to activated cryptic splice-site was not statistically different between TP and FP (two-sided Wilcoxon rank sum test: W = 2,834, p = 0.70, n = 124 Top-4* FP and 44 TP biologically independent cryptic-activation events). A-E and H-I are box-whisker plots, with internal lines denoting the median value, and the lower and upper limits of the boxes representing 25^th and 75^th percentiles. Whiskers extend to the largest and smallest values at most 1.5IQR. * = skipping of one or two exons or cryptic activation within 600 nt of the annotated splice-site.

Source data