PRO-seq and RNA-seq data preparation and processing

Our main analysis is based on PRO-seq data for K562 [28] and HeLa [70] cell lines as well as RNA-seq data from the ENCODE project [48, 71] (ENCSR000AEM for K562, ENCSR000CPR for HeLa). For comparison, we also sequenced new PRO-seq (n = 2) and RNA-seq (n = 4) libraries, generated from cells grown in the same flask under the same conditions. Human K562 cells were cultured using standard cell culture procedures and sterile techniques. The cells were cultured in RPMI-1640 media supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. For PRO-seq, 3′ and 5′ adapters were ligated as described [26] followed by library preparation as previously published [72]. Sequencing was done by Novogene on a HiSeq instrument with paired-end reads of 2 × 150 bp. For RNA-seq, total RNA was extracted using the Trizol method (see https://assets.thermofisher.com/TFS-Assets/LSG/manuals/trizol_reagent.pdf), followed by rRNA depletion using the Ribozero HMR Gold kit. Libraries were prepared using the NEB kit with TruSeq RNAseq adaptors. Single-end sequencing (length = 75) was performed on a NextSeq500 instrument by the RNA Sequencing Core at the College of Veterinary Medicine, Cornell University. Sequencing data is deposited on GEO under accession {"type":"entrez-geo","attrs":{"text":"GSE153200","term_id":"153200"}}GSE153200.