phyloDS: phylogeny-aware differential splicing across species, beyond reference annotations

Alternative splicing varies widely between species, but its comparative analysis has lagged significantly behind that of gene expression analysis. Traditional methods define junction orthology using a single reference genome, treat species as independent samples, and rarely rely on sequence-level evidence to substantiate detected shifts. To address these limitations, researchers have developed phyloDS, an RNA-seq-driven framework.

The phyloDS approach identifies orthologous splice junctions by analyzing multiple sequence alignments of one-to-one orthologous genes. It then tests the differential usage of these junctions with a regularized Ornstein-Uhlenbeck model. Crucially, phyloDS cross-validates its findings against splice-site predictions generated solely from genomic sequence data, providing a more robust analysis.

When applied to primate brain and rodent heart transcriptomes, phyloDS demonstrated its effectiveness. It uncovered up to 2.9 times more shared orthologous splice junctions compared to a reference-based strategy, maintaining type I error rates close to nominal values where competing methods inflated this rate by over two-fold. Furthermore, phyloDS identified 234 and 502 differently used orthologous splice junctions in hominoids and mole-rats, respectively.

These findings include 86 and 141 previously unannotated junctions, revealing unique mitochondrial and contractile splicing signatures in mole-rats.