Enhancer-gene interactions called with the generalised Activity-by-Contact score on the IHEC H3K27ac ChIP-seq data (https://doi.org/10.1093/bioinformatics/btad062). All files are in hg38. As gene annotation the v38 gtf-file from GENCODE was used. The samples in each of the files refer to all observed H3K27ac ChIP-seq experiments, according to the sample metadata version 1.1 (https://github.com/IHEC/epiATLAS-metadata-harmonization/blob/main/openrefine/v1.1/IHEC_metadata_harmonization.v1.1.csv) and the experiment metadata from the sftp (downloaded 23th July 2023). ------------------------------------ V1_PeakBased_gABC/ ------------------------------------ Peak-gene interactions called with the gABC-score with two output files for each H3K27ac ChIP-seq sample (n=1565). The H3K27ac ChIP-seq peaks called with MACS2 by IHEC were used as candidate enhancer for each sample. The MACS2 signalValue was used as enhancer activity. As contact data the average Hi-C matrix from https://doi.org/10.1101/2023.11.09.563812 was taken. Window-size was set to 5 MB, the score cutoff to 0.02, and all regions intersecting the ENCODE blocklisted regions were excluded (https://github.com/Boyle-Lab/Blacklist/tree/master). The command looked as follows and was repeated for each MACS2 narrowPeak file: STARE_ABCpp -b -n 7 -i all_tss -a gencode.v38.annotation.gtf -w 5000000 -t 0.02 -f ENCFF134PUN_avgHiC_hg38/ -k 5000 -x hg38-blacklist.v2.bed -o ------------------------------------ V1_PeakBased_gABC_UCSCtracks/ ------------------------------------ The matching UCSC interact files for each of the interaction files in the V1_PeakBased/ directory, which can be loaded into the UCSC genome browser. Due to the large window size, the visualization can be tricky. We recommend trying out the 'shrink' and 'pack' display modes. ------------------------------------ V2_StackedChromHMM_gABC ------------------------------------ Files based on gABC interactions using merged bins from the stackedChromHMM_V1 as candidate enhancers (see below). V2_StackedChromHMM_EnhancerStates_K27FC.bed.gz: A matrix of regions*samples. Bins assigned to one of the following states by the stacked ChromHMM (version 1) were merged into a set of regions: 'EnhWk', 'EnhA', 'TxEnh', 'BivProm', 'PromF', 'TSS', 'Acet'. For each of the resulting merged regions, their sample-specific H3K27ac signal (average fold-change over background) was collected. For each sample/column, entries where the region does not overlap a sample-specific MACS2 H3K27ac ChIP-seq peak were set to zero. V2_gABC_StackedChromHMM_InteractionsSampleMat.txt.gz: Matrix of interactions * samples with the gABC score as values. Scores ≤0.02 are set to 0. Candidate enhancers and their activity for the scoring are the ones from the file above (V2_StackedChromHMM_EnhancerStates_K27FC.bed.gz). As contact data the average Hi-C matrix from https://doi.org/10.1101/2023.11.09.563812 was used. Window-size was set to 5 MB, the score cutoff to 0.02, and all regions intersecting the ENCODE blocklisted regions were excluded (https://github.com/Boyle-Lab/Blacklist/tree/master). The command looked as follows: /projects/triangulate/work/STARE/Dev_Code/STARE_ABCpp -b V2_StackedChromHMM_EnhancerStates_K27FC.bed -n 4+ -i all_tss -a gencode.v38.annotation.gtf -w 5000000 -t 0.02 -f ENCFF134PUN_avgHiC_hg38/ -k 5000 -x hg38-blacklist.v2.bed -o V2_gABC_StackedChromHMM_InteractionsSampleMat_Bin.txt.gz: Same as above but the entries are binarized, meaning everything >0 is set to 1. V2_gABC_GeneSummedEnhancer.txt.gz: Matrix of genes * samples. For each gene sums up the H3K27ac signal of all enhancers linked to the gene in the sample according to gABC_StackedChromHMM_InteractionsSampleMat.txt.gz. Genes without any non-zero value across all samples are removed. V2_gABC_GeneSummedEnhancer_geq2.txt.gz: Same as above, but sets the values to 0 where the maximum promoter signal (H3K27ac FC in ±200bp around all annotated TSS) of the gene was < 2. Contact for questions: Dennis Hecker, d.hecker@em.uni-frankfurt.de