Differential analysis of PRO-seq

Differential analysis was performed using a non-redundant set of genes with CAGE corrected TSSs. For each gene the region from the TSS up to 150 bp downstream (+1 to +150) was defined as “promoter + pause region”, and the rest of the annotated gene was defined as “gene body”. For BRD4 depletion in BRD4-AID cells (Fig. 4f,g), the number of unique (UMI collapsed) PRO-seq read 5’ ends falling into these two regions was counted for each gene. Differential analysis was performed with DESeq2 v.1.22.2 (ref. ⁷⁴) for “promoter + pause” and “gene body” region separately to capture the pause-release defect. For MED14 depletion in MED14-AID cells and induction of P53 target genes by Nutlin-3a in WT, MED14- and BRD4-AID cells (Fig. 3d,e, Extended Data Fig. 4a,d), number of unique (UMI collapsed) PRO-seq read 5’ ends falling into the whole gene region was counted and differential analysis was performed on the entire gene. Raw PRO-seq counts used for differential analysis are provided in Supplementary Table 6. To allow accurate assessment of changes in enhancer activity upon different treatments and possible detection of global effects, we used spike-in based normalization. Scaling factor for normalization between conditions was calculated from relative abundance of reads mapping to spike-in genome (dm3) in combined replicates for each condition. Spike-in normalization factors were supplied as custom scaling factors to DESeq2, with all replicates of the same condition receiving the same scaling factor. These scaling factors were also used to normalize PRO-seq coverage of combined replicates per condition for visualization in the genome browser. Spike-in read counts, relative abundances, and calculations of scaling factors are provided in Supplementary Table 5 – PRO-seq (MED14-, BRD4-AID & WT).