The signal files were downloaded from the ENCODE portal, taking only the plus and minus strand signal of unique reads belonging to each experiment's default analysis. ENCODE publishes no pooled file, so the per-replicate files were summed at UCSC into one track per cell line and strand, the same merge the ProCapNet models were trained on. Total signal is conserved exactly by the summing, and minus strand signal is negative as released. The experiments are ENCSR046BCI, ENCSR100LIJ, ENCSR935RNW, ENCSR098LLB, ENCSR261KBX and ENCSR799DGV for A673, Caco-2, Calu3, HUVEC, K562 and MCF10A respectively.
The steps are recorded in doc/hg38/transcriptionStart.txt and the scripts are in +target="_blank">doc/hg38/transcriptionStart.txt, the scripts are in makeDb/outside/proCapNet, including proCapNetEncodeFiles.tsv, which records exactly which ENCODE file -accessions went into each track. +accessions went into each track, and the track configuration is in +trackDb/human/hg38/transcriptionStart.ra.
The bigWig files are on our download server, named for the cell line, the ENCODE experiment accession and the strand, for example K562.ENCSR261KBX.pos.bw and K562.ENCSR261KBX.neg.bw.
The data can be explored interactively in table format with the Table Browser or the