Detect trend among multiple ChIP-seq data

Question

[dcheng1]

I want to find loci where peak signal change gradually among multiple ChIP-seq data. For example, in the picture below, peak intensity gradually decrease from the top data to the bottom. I'm wondering if there are any tools can help me find those loci? Any suggestions are greatly appreciated!

[shoujun.gu]

I'm afraid you have to write your own code to do this work, such as: bin the region-> calculate mean peak value -> filter non-decreasing region.

[dcheng1]

Thanks for your help!

[ATpoint]

I would use heatmaps/hierarchical clustering to tackle this problem. Call peaks or bin the genome into regions of something like 1kb, combine peaks into a consensus list. Quantify reads over this list (e.g. with featureCounts), normalize data (e.g. vst from DESeq2) and then make a heatmap e.g. with ComplexHeatmap and select the cluster(s) that matches the desired pattern.

[dcheng1]

I thinks I understand the part before heatmap. Assume I have n peaks in the consensus list, and I have m data. Then I will have a n x m matrix. Each value in the matric is the normalized read counts. Then I guess I need to plot a heatmap using the n x m matrix? I don't understand "select the cluster(s) that matches the desired pattern". Does that mean I need to change cluster method?