Alternative splicing regulatory network reconstruction from exon array data.

Pre-mRNA alternative splicing (AS) allows individual genes to produce multiple types of mRNA and associated protein isoforms. While AS regulation enables the production of the hundreds of thousands of types of proteins needed for the normal functioning of the human cell, it also presents many opportunities for the onset of cancer and other diseases. The AS process is known to be regulated by a group of serine/arginine rich (SR) proteins, heterogeneous nuclear ribonucleoproteins (hnRNPs), and small nuclear ribonucleoprotein (snRNP) particles through a complex assembly. Each gene-exon is regulated by one or multiple splicing regulators, from which one may hypothesize the existence of an alternative splicing regulatory network (SRN). The SRN contains a list of gene-exons, for each of which the factors that up/down regulate them are provided. Since defects in the SRN play key roles in human disease, a reconstruction of human SRN could be used to facilitate the design of diagnostic and therapeutic strategies. In this paper, we present a methodology to automate genome-wide SRN reconstruction. We construct SRN based on an extensive correlation analysis of human exon expression microarray data, conventional gene expression microarray profiles, and an experimentally verified AS and transcriptional regulatory interaction training set. This SRN reconstruction methodology is demonstrated and software (AutoNet) that automates the reconstruction of SRN is developed. A genome-wide SRN was constructed for normal human cells and an assessment of the reliability of each predicted interaction is provided. Human SRN we constructed are free available from our web portal: https://ruby.chem.indiana.edu/scorenfl/srn_results/lookup0.php.