Large-scale RNA-seq analysis identifies over 3,000 recursive splice sites in human introns

A preprint posted to bioRxiv presents a new computational approach for detecting recursive splice sites (RSSs) and applies it to a large total RNA-sequencing dataset from human brain tissue, identifying 3,022 RSSs distributed across 2,775 introns in 2,407 genes.

Recursive splicing (RS) is a process in which a long intron is excised from a pre-messenger RNA in two or more sequential steps rather than in a single reaction. It was first characterised in Drosophila and has been documented in mammals, but its full prevalence in the human transcriptome has been difficult to assess because standard RNA-seq pipelines are not optimised to detect it.

The authors developed a detection strategy based on aligning RNA-seq reads to short, customised target sequences designed to capture RS intermediates, using parameters calibrated to achieve a very low false-discovery rate. Applying this to brain expression data, they found that 96% of detected RSSs — some 2,891 sites — appeared in their analysis, suggesting the phenomenon is a routine feature of transcription across thousands of human genes rather than an exceptional case.

The findings have potential relevance to understanding splicing regulation, disease-causing splicing variants, and the interpretation of long-intron genes. The preprint does not report disease associations but notes that many affected genes are highly expressed in the brain. This work has not yet undergone peer review.