Allo-only and UMR peak calls for 481 K562 datasets

We analyzed the ENCODE public database, focusing specifically on ChIP-seq experiments in K562 cells. From 774 qualifying experiments (retrieval date Feb. 2023), we selected 481 based on the criteria that the dataset had at least one replicate, had a control file, no major audits from ENCODE, and resulted in non-zero peak calls in the UMR sample using our pipeline. We also only analyzed one dataset per transcription factor, which was randomly selected from the selection of those available. We retrieved FASTQ files for both single- and paired-end experiments. Replicates were concatenated and aligned using Bowtie 2 v.2.5.1, reporting 25 valid alignments per read with parameters “--no-mixed --no-discordant” for paired-end reads. The alignments were sorted using SAMamtools v1.16.1 collate. We employed Allo on the sorted alignments to obtain rescued reads. For control experiments, reads were randomly assigned during Allo processing (--random). Following alignment and rescue, peaks were subsequently identified using MACS2 v2.7.1, with the argument “-f BAMPE” for paired-end experiments. ENCODE blacklist regions were excluded using BedtoolsBEDTools v2.27.1 intersect -v.