Observations on novel splice junctions from RNA sequencing data

High-throughput RNA sequencing (RNA-seq) technology provides a revolutionary approach to studying splicing events de novo. However, identifying splice junctions with high sensitivity and specificity remains a challenge. In the present study, we proposed a new tool named SeqSaw to detect splice junctions with or without the canonical GT–AG splicing signal. SeqSaw was applied to two ENCODE RNA-seq datasets and also compared with two existing methods. It was shown that the proposed method obtained better results on finding novel splice junctions. Experiments also revealed that the current sequencing depth has not yet reached saturation to detect novel transcripts. Moreover, by comparing the number of supporting reads, we demonstrated that many un-annotated splicing events can be tissue specific.

► A proposed method efficiently detects novel junctions from RNA-seq data.
► Detected junctions indicate many previously unknown splicing events.
► Effect of sequencing depth on junction detection varies with junction type.
► Most of the detected novel junctions belong to known gene regions.
► Many of the detected novel junctions show tissue-specific expression.