Genrich analyse ATAC-seq data problem

Question

[MatthewP]

I am trying to analyse ATAC-seq data with Genrich with 2 replicates. I want to ask a few questions.

1. Output file(narrowPeak) by -o is peaks of all samples or is peaks that different between 2 groups?
2. I need to know some genes' promoter region have different peak or not? Will peak annotation give the result?
3. If I find some different peaks in promoter region and I want to visualize all samples' peak at that region, anytools good？I can find Genrich -b will output bed files contain read name of intervals, this seems to use this output.

Thanks.

Command:

Genrich -t ${BamDir}/KO1_ATAC.bam,${BamDir}/KO2_ATAC.bam -o ${GenrichDir}/ATAC.narrowPeak -c ${BamDir}/WT1_ATAC.bam,${BamDir}/WT2_ATAC.bam -f ${GenrichDir}/ATAC_pq.bed -k ${GenrichDir}/ATAC_pileup_p.bed -b ${GenrichDir}/ATAC_reads.bed -r -e MT -E ${Blacklist} -m 30 -j

[ATpoint]

1. Output file(narrowPeak) by -o is peaks of all samples or is peaks that different between 2 groups?

With the code below it would be the peaks that are significant considering the two KOs as replicates and then WTs as control. Genrich is replicate-aware, please check the manual. I would be skectical though to use an experimental condition as control, this is not the idea of the -c parameter. It is rather to correct for spurious background enrichment. In ChIP-seq that would be an input or IgG control which you don't have. I would rather call peaks on the KOs and WTs separately, then merge the peaks, make a count matrix and analyse with edgeR or DESeq2 (or any other diff. testing framework) to call differential regions.

1. I need to know some genes' promoter region have different peak or not? Will peak annotation give the result?

Intersect the differential results from 1) with promoter coordinates. Please browse biostars on how to get promoter coordinates. It is basically the upstream regions of annotated TSS.

1. If I find some different peaks(...)

Maybe a heatmap like Using EnrichedHeatmap for visualization of NGS experiments ?

[MatthewP]

Ok, here you mean -c is sample without transposase treatment instead of experiment design control sample? I will run Genrich again separately and also try MACS2, thanks.

[ATpoint]

No, it would technically be genomic DNA with transposase treatment but no one is actually doing this as it would reqiure quite some sequencing depth to actually get enough signal across the entire genome to serve as a proper background. Just leave -c out for ATAC-seq. Your control samples will become important during the differential analysis, not during peak calling.

[MatthewP]

Just assume that I want to prepare this -c control sample, what's the experiment different between this control sample and my experiment design control sample(Cell without special treatment)?

[ATpoint]

As I said above, the proper control would be genomic DNA digested with the Tn, so the entire genome irrespective of open- and closed chromatin to get an idea of the Tn insertion preferences, amplification bias etc.