Identify genes with chip signal > input prior to peak calling

Question

[SciDuck]

Hi everyone,

I need to identify which genes in my histone ChIP-seq experiment have greater signal for chip than for input (enrichment). I have made an enrichment plot deepTools plotProfile, however, the profile contains negative values which I suspect are a result of regions not enriched for the histone mark of interest. This is okay but I also want to be able to plot only the regions with enrichment.

I know one way to do this would be with peak calling. But given that peak calling doesn't call all the enriched regions and since I am looking to plot the average enrichment profile of the histone mark over the all the TSS in the genome, I do not want to exclude any regions with enrichment. So I would like to find a way to identify which regions in the genome have chip signal>input and then use this for plotProfile.

Does anyone know of a way to do this besides peak calling? Please help!

[Gautier Richard]

I think using clustering with plotHeatmap will yield what you want. So either --kmeans or --hclust : https://deeptools.readthedocs.io/en/develop/content/tools/plotHeatmap.html

[SciDuck]

Thank you so much, Gautier! This is exactly what I needed. Much appreciated!

[Prakash]

what I understood that you want to narrow down ChIPseq signal having enrichment > input (but how much 1.5 fold or 2 fold?), and then use look for enrichment near TSS. I think you need keep one thing fixed either all the TSS or ChIPseq enrichment. I would suggest to to reduce the fold change cutoff while calling peak.

[SciDuck]

Thank you Prakash, I will try reducing the fold change during peak calling.