ChIP-seq data processing

Clean reads, obtained by removing adapter sequences from the raw ChIP-seq data with custom scripts, were mapped to the reference genome [University of California, Santa Cruz (UCSC) human hg19 or the custom-combined reference genome] using Bowtie2 (version 2.2.3) with default parameters.

For H3K14ac, clean reads were mapped to the custom-combined reference genome, which was concatenated with the human (UCSC hg19) and Drosophila (UCSC dm6) genomes, as previously reported (63). To build the combined reference genome, we labeled chromosome names in the Drosophila reference genome with the ‘_dm6’ suffix so that we can easily separate the reads mapped into the human or Drosophila furtherly. A custom alignment library was built for the combined reference with Bowtie2. After the mapping process, only uniquely mapped nonduplicate reads were retained with MACS2 (version 2.1.1.20160309). To quantify the ChIP-seq signal, a normalization factor was used as previously reported (63). Briefly, we defined α as the normalization factor, β as the signal from the Drosophila cells, Nd as the number of reads (in millions) uniquely mapped to the Drosophila reference genome and r as the percentage of Drosophila cells. Therefore, the formula was defined as followsβ=α×Ndγ

Because the signal from Drosophila cells β and the percentage of Drosophila cells r was constant across samples, we simplified these values to β = 1 and r = 1. Accordingly, the normalization factor was defined as followsα=1/Nd

The ChIP-seq signals were normalized to the normalization factor and visualized with software deepTools.

For H3K4me3, we directly mapped the ChIP-seq data to the human genome hg19 and normalized the ChIP-seq signals to the number of reads uniquely mapped to the human reference genome.