Chip Seq analysis for H3K27me3,H3K27ac and K4me1 enrichment

Question

[Researcher]

Hi, I am new to chips analysis. My aim is to compare the histone marks from two different subtypes of a cancer and want to further integrate these results with the transcriptomic information from the same samples.

I am analyzing my chipseq data enriched for histone marks by using following steps:

1. Quality filtering of raw reads,
2. Mapping with bowtie,
3. Converting to bam, indexing followed by sorting using samtools
4. Preprocessing of the aligned reads using deep tools in the following manner

a) Correlation between bam b) Coverage check c) GC bias check and correction

1. GC corrected bam was used to generate matrix with reference point using computeMatrix which was followed by plotting the heat map.

In order to do this I have few queries:

A) While using compute matrix from deep tools I am getting so many peaky around the TSS and TES, but normally it should be a sharp peak. Please correct me if I am doing something wrong.

B) How can I integrate MACS2 to call peaks with the above said pipeline.

C) What should be the best practice and how should I proceed to find potential differentially regulated genes or regions.

Any suggestions and help will be highly appreciated.

Thanks

[ATpoint]

Did you do any literature or internet search before? You cannot expect that people provide you a hand-written workflow. Look at the vignette of csaw for inspiration on ChIP-seq analysis. H3K27me3 by the way is a broad rather than a sharp histone modification. Please invest time into getting a background then then come back with more specific questions.

[Researcher]

Hi ATpoint, Thanks for your recommendation I ll go through the reference. As I am new to R too, thus was looking for an user friendly approach to address this question.

Answer 1

[jared.andrews07]

As for peaks near/around the TSS, distal enhancers will also frequently have H3K27ac and H3K4me1 present, so don't be surprised to see peaks at other regions or that signal around the TSS spreads a bit more than you would expect.

As for peakcalling (which isn't even necessary with csaw, MACS could be used with your GC corrected bams to call peaks.

diffBind and csaw are probably the most common and easiest to use tools for differential binding. Tying differentially bound regions to genes can be a trickier task, though there are tools out there that attempt that as well (GREAT comes to mind, though there are others as well). Changes at the TSS are easy to handle, but intergenic regions are much tougher.

[Researcher]

Thank you jared for your kind reply.