Quantification of co-transcriptional splicing from RNA-Seq data

During gene expression, protein-coding transcripts are shaped by multiple processing events: 5' end capping, pre-mRNA splicing, RNA editing, and 3' end cleavage and polyadenylation. These events are required to produce mature mRNA, which can be subsequently translated. Nearly all of these RNA processing steps occur during transcription, while the nascent RNA is still attached to the DNA template by RNA polymerase II (i.e. co-transcriptionally). Polyadenylation occurs after 3' end cleavage or post-transcriptionally. Pre-mRNA splicing - the removal of introns and ligation of exons - can be initiated and concluded co-transcriptionally, although this is not strictly required. Recently, a number of studies using global methods have shown that the majority of splicing is co-transcriptional, yet not all published studies agree in their conclusions. Short read sequencing of RNA (RNA-Seq) is the prevailing approach to measuring splicing levels in nascent RNA, mRNA or total RNA. Here, we compare four different strategies for analyzing and quantifying co-transcriptional splicing. To do so, we reanalyze two nascent RNA-Seq datasets of the same species, but different cell type and RNA isolation procedure. Average co-transcriptional splicing values calculated on a per intron basis are similar, independent of the strategy used. We emphasize the technical requirements for identifying co-transcriptional splicing events with high confidence, e.g. how to calculate co-transcriptional splicing from nascent RNA- versus mRNA-Seq data, the number of biological replicates needed, depletion of polyA+RNA, and appropriate normalization. Finally, we present guidelines for planning a nascent RNA-Seq experiment.