A machine learning strategy to identify candidate binding sites in human protein-coding sequence

This strategy identified a collection of motifs including several previously reported splice enhancer elements. Although only trained on coding exons, the model discriminates both coding and non-coding exons from intragenic sequence.We have trained a computational model able to detect signals in coding exons which seem to be orthogonal to the sequences' primary function of coding for proteins. We believe that many of the motifs detected here represent binding sites for both previously unrecognized proteins which influence RNA splicing as well as other regulatory elements.In eukaryotes, the majority of transcripts are processed by splicing to remove intron sequences and form a mature messenger RNA. There are well established conserved sequence motifs at the intron/exon and exon/intron boundaries that are known to be recognised by the splicing machinery. However, even for short intron sequences, it has been concluded that these signals do not contain sufficient information to accurately define splice sites [1] and that other splicing factors, with associated binding sites, must be involved.There is evidence that splicing is partly promoted and regulated by sequences embedded within exons. A number of sequences have been found embedded in the exons of both viral and cellular genes which can promote or repress the utilization of alternative splice sites. These are usually purine-rich sequence located near an alternative splice donor, that bind splicing factors such as members of the SR family. SR proteins are highly conserved serine/arginine-rich RNA-binding proteins (For review, see [2]). They are essential splicing factors and have been shown to regulate the selection and use of alternative splice sites [3-10]. It is known that they function very early in the spliceosome assembly process, promoting the binding of U1 snRNP to the splice donor and of U2AF to the splice acceptor, apparently by interacting with U1 70 K and U2AF respectively. Observations have shown that S