Flowchart describing the steps used by RNAmotifs to identify the enriched multivalent RNA motifs. The multivalent RNA motifs are predicted by assessing clusters of tetramers that are enriched in the genomic sequence at specific positions relative to enhanced or silenced exons, compared to control exons. Clusters of each tetramer are evaluated in three regions around the splice sites of alternative exons. Analysis of control exons is used to determine the clustering threshold that each tetramer needs to reach before it is considered as a ‘cluster instance’. A one-tailed Fisher’s exact test is then used to test the null hypothesis that the number of cluster instances at a precise region of a particular tetramer is not different between enhanced (or silenced) and control exons, and the Benjamini-Hochberg false discovery rate (FDR) correction is applied to calculate p
. For each tetramer, the achieved significance level of the test (pempirical) is calculated with a bootstrap procedure using 10,000 samples. Tetramers with p
≤0.1 and pempirical ≤0.0005 in at least one region either in the enhanced or silenced set are retained (Additional file 2). The RNA map is then drawn to visualize the enrichment score at each nucleotide around the enhanced or silenced exons, and their flanking exons. nts, nucleotides; ss, splice site.