Fig. 4

Co-splicing network construction. The following steps were performed to construct a co-splicing network: First, the sets of transcripts highly correlating with at least one SRP (Spearman correlation coefficient > 0.95) were identified. Second, 30-nucleotide regions surrounding each exon/intron splice junction (R1 through R4) were extracted for each transcript co-expressing with any of the selected RBPs. In this example, transcripts X and Y are encoded by genes X and Y. Third, the resulting sequences were subjected to MEME analysis separately for each of the SF groups to find consensus motifs. Fourth, transcripts containing significantly enriched motifs at each region (p-value < 0.05) were retrieved from the MEME results. Fifth, for each transcript that had a significantly enriched motif in at least one of the R regions (e.g., R1 and R3 for co-expressing genes X and Y), the corresponding weighted edges representing the individual R regions were constructed between each SF and its potential product in Cytoscape. Edges are weighted based on –log (p-value).