Abstract
RNA-binding proteins (RBPs) recognize consensus sequences and regulate specific target mRNAs. However, large-scale CLIP-seq revealed loose and broad binding of RBPs to larger proportion of expressed mRNAs: e.g. SFPQ binds to >10,000 pre-mRNAs but distinctly regulates <200 target genes during neuronal development. Identification of crucial binding for regulation and rules of target recognition is highly anticipated for systemic understanding of RBP regulations. For a breakthrough solution, we developed a bioinformatical method for CLIP-seq and transcriptome data by adopting iterative hypothesis testing. Essential binding was successfully identified in C-rich sequences close to the 5' splice sites of long introns, which further proposed target recognition mechanism via association between SFPQ and splicing factors during spliceosome assembly. The identified features of functional binding enabled us to predict regulons and also target genes in different species. This multilateral bioinformatics approach facilitates the elucidation of functionality, regulatory mechanism, and regulatory networks of RBPs.