Fig. 4: Impact of sQTLs on splice sites and RBP binding sites.
a Distribution of the absolute change in splice site strength for sQTLs with low, moderate and high effect sizes (MD value). b Distribution of the absolute deltaSVM value (∣deltaSVM∣) of sQTLs and non-sQTLs, for RBPs with significantly different mean ∣deltaSVM∣ between sQTLs and non-sQTLs (two-sided Wilcoxon Rank-Sum test, FDR < 0.05, total sample size for each test listed as follows: nRBFOX2 = 509, nPRPF8 = 1,133, nGTF2F1 = 376, nSF3B4 = 595, nGRWD1 = 736, nPPIG = 691, nGEMIN5 = 345, nCSTF2T = 1,170, nRBM15 = 373 and nXRN2 = 450). Data is shown as boxplots, where the box represents the first to third quartiles and the median, and the whiskers indicate ± 1.5 × interquartile range (IQR). c Modification of the binding sites of the RBPs RBFOX2 (left) and PRPF8 (right) by SNPs rs4959783 (chr6:3,260,093, G/A, ∣deltaSVM∣ = 2.48) and rs9876026 (chr3:11,849,807, T/G in the reverse strand, ∣deltaSVM∣ = 4.77), respectively. The lines represent the gkm-SVM scores of all possible (overlapping) 10-mers in a 100 bp window around the SNP. Those corresponding to the 10-mers overlapping the SNP are colored according to the allele. SNP positions are marked with a dashed line. The gray area includes the 90% middle gkm-SVM scores of 10-mers not overlapping the variant. The relative location of the predicted RBP motifs and the corresponding sequence logos are also displayed. In the logos, the SNP position is marked with an asterisk. Source data for a–c are provided as a Source Data file.
