4.12.3. ChIP-Seq Analysis

FASTQ files were controlled for quality issues, using FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Read alignment against the hg19 human genome reference was downloaded as a pre-compiled BWT index from Illumina’s iGenome repository (https://emea.support.illumina.com/sequencing/, sequencing_software/igenome.html). Read alignment was performed by using bowtie version 1.1.2 with parameters -k 1 -m 1. Duplicate removal was performed by using Picard’s MarkDuplicates and Samtools rmdup function. Coverage vectors were generated with Deeptools bamCoverage function, using RPKM (reads per kilo base per million mapped reads) normalization [87]. Visualization of binding profiles was done by using the R/BioConductor package Gviz. H2A.Z and H2A.Zac peak calling was done using MACS2 [88]. The resulting set was filtered against blacklisted chromatin regions, as detected by ENCODE. Differential binding analysis was performed by using DESeq2 [81] after merge peaks into reference peak sets, using reduce function. In order to identify overrepresented functional terms amongst binding-site-associated genes, we used GREAT (Genomic Regions Enrichment of Annotations Tool) [89], using standard settings. For heatmap representation of binding data across binding sites, coverage vectors were collected in defined intervals across binding sites or RefsSeq TSSs, using Deeptools computeMatrix, and plotted via Deeptools PlotHeatmap function. The comparison to chromatin states was performed as previously described [55].