“…We calculate the splicing index value θ (Mukherjee et al, 2017) (Supplementary Figure 2A) to determine splicing efficiency across all time points for introns that have at least 5 reads coverage on both 5’ and 3’ SJ for all included RNA sequencing libraries (four time points of BrU-pulse-chase-seq and the three Input samples for TNT-seq), yielding 13,532 introns with an extracted θ value ranging from 0 (unspliced) to 1 (fully spliced) (Methods). As expected, the cumulative distribution of the splicing index at 0 min, representing nascent RNA, precedes that of the steady-state chromatin-associated RNA (Conrad et al, 2014) indicating that pre-mRNA is efficiently captured by the assay (Figure 2A). Using k-means clustering with k = 3 we called three clusters of distinct splicing efficiency dynamics (SED) representing 4,882 fast, 5,702 medium, and 2,948 slowly processed introns (Figure 2B and Supplementary Figure 2B-D) (for the definition of SED see Methods under ‘Splicing kinetics and predictive models’).…”