Contributing to proteomic diversity, alternative splicing of
pre-mRNA is widespread in the human transcriptome and can greatly influence regulation of both normal and disease-related cellular phenotypes. Similar to gene expression, alternative splicing does not occur independently, but in a coordinated fashion throughout the transcriptome in order to maintain proper cellular function. Gene co-expression networks have been widely used as an approach to elucidate coordinated regulatory patterns of gene transcription. Studies have shown that genome-wide expression can occur in the form of network modules consisting of highly co-expressed genes operating within specific cellular pathways. Such modules are often well-preserved across similar biological systems and associated with various phenotypes. Here we demonstrate a framework for de novo network inference of co-splicing in the form of modules consisting of complex alternative splicing variants.