Peak density heatmap

Question

[Amlan]

I am not sure, if the title of this post is appropriate or not. I have some ChIP-Seq data (MACS peaks) for different fat depots with H3K4Me3 signal. I want to show the comparison of the density and distribution of the peaks. So, I actually need an image in the link below. I don't know if it will be called a peak density heatmap, as I could not get enough help on the internet using that as search key. Do anyone have any experience on making such kind of heatmaps?

I have made some heatmaps in Python. So, some help on Python would be great, although any suggestion is welcome.

http://www.bloodjournal.org/content/bloodjournal/124/25/3719/F3.large.jpg?sso-checked=true

Thanks in advance, Amlan

[Sinji]

These look like typical heatmaps. I suggest taking a look at this post.

Answer 1

[Sukhi Singh]

Check the post from Sinji Visualization of ChIP-seq data using Heatmaps (Updated: 03/10/16)

Other tools based on Python capable of generating coverage plots and heatmaps are MetaSeq and HtSeq

Reference : metaseq: a Python package for integrative genome-wide analysis reveals relationships between chromatin insulators and associated nuclear mRNA

Answer 2

[igor]

It looks like the plots in that figure are made by ngs.plot: https://github.com/shenlab-sinai/ngsplot

Answer 3

[Amlan]

Thank you a lot. That link is very helpful to make heatmaps. But it does not serve my purpose. May be the link I posted was misleading. Sorry about that.

Actually I need to compare the density and genomic distribution of two peak files. I have two H3K4Me3 peak files for two different fat depots. Now I need to show the difference of those peaks based on their genomic positions and at the same time I also need to show their peak density. So this should be something like the following image. Here Eif01A and Eif01G are the two H3K4Me3 peak data for the two fat depots and the image shows the comparison of their genomic distribution i.e. which genomic positions have both signals and which genomic positions have only one of them.

[Sinji]

This is a typical heatmap with K-means clustering it looks like. I suggest you take a look at deepTools as it should be able to do what you need. You'll need to start off with a computeMatrix step, followed up by a plotHeatmap step where you would use k-means clustering to get these groupings.