Detection of novel splice forms in human and mouse using cross-species approach

Millions of transcript sequences have become available for characterizing the transcriptome of human and mouse. Transcript databases have been extensively mined for extracting alternative splicing information within the same species; but they also represent a potentially valuable resource for the discovery of alternative splice variants in another species. In this study, we have performed analysis of alternative splicing patterns for 7,475 pairs of human and mouse genes. We found that cross-species transcript analysis could accomplish the same level of sensitivity in detecting constitutive splice patterns as EST resource from the same species. In contrast, identifying alternative splice patterns in human genes, mouse transcripts achieved only 50% of the sensitivity of human EST and 70% of the sensitivity of human mRNA. While identifying alternative splice patterns in mouse genes, human transcripts are 38% more sensitive than mouse mRNA, and reach 60% of the sensitivity of mouse EST. Furthermore, using the cross-species approach, we predicted novel alternative splice patterns for 42% of human genes and 51% of mouse genes. Splice site motif analysis suggests that the majority of predicted novel splice patterns are expressed in human. EST-based frequency analysis shows that novel splice patterns are expressed at lower frequency than alternative splice patterns present in the transcript data from both species, possibly explaining why they remain undetected in the transcript data of the same species.