Figure 3.

RBP motif enrichment analysis. A: PCA of motif-derived k-mer z scores, with each point as a motif and the vector components as the regulation of splicing (inclusion or exclusion). Motifs with PCA distance >2.5 SDs away from zero are plotted in red and labeled with the motif ID and RBP name from CISBP (version 6.0). Enriched motif sequences in each splicing regulation are conserved with distinct nucleotides. B: Heatmap showing differential mRNA expression levels of RBPs with PCA distance >2.5 SDs from zero. The heatmap represents the z score of RPKM values, with each row representing a gene. Each column represents a sample: Cont indicates human islets without cytokine treatment, and Cyto indicates human islets with cytokine treatment, ordered the same as the Cont group. C: Pairwise correlation between the 16 binding motifs for all 12 differentially expressed RBPs. D: Shown is the network of RBPs and the genes harboring splicing events. Nodes in the network correspond to the RBP and significant splicing exons with an ExonImpact prediction score ≥0.5. The color of the node represents differential splicing ratios. Edges represent the prediction count by FIMO, which measures the probability that an RBP binds to a target gene. The line thickness increases with number of predicted counts as indicated. Node sizes increase with significance (−log₁₀ FDR) of the splicing events as indicated. The left panel depicts the network for all 12 RBPs, and the right panel depicts the SRSF2 network only.