Peak calling for broad histone-modification regions

Question

[biostart]

Hello,

Is someone finding broad histone modification regions using peak callers such as MACS or HOMER? They only find narrow 1-2kb peaks for me, while expecting to find broader peaks such as 10-100kb. Any suggestions?

Thanks

[Ming Tommy Tang]

take a look at epic https://github.com/biocore-ntnu/epic

[ivivek_ngs]

How can you say that you expect to have such broader peaks for your data? How did you call the broad peaks and which are the histone modifications are you talking about when you say broad peaks. Show us the parameters you have used and for which histone marks and then we can say. I can mostly help with MACS2 and unless I see your command line I will not be able to say if it was correctly done or not. As for your query, are you seeing the same range for both MACS2 and HOMER? Are you using MACS2(latest version)? Have your created the bedgraph of the normalized broad output of the MACS2 and viewed it in a browser against the ones called with narrow since not all histone marks should be called with broad parameters. If you can provide these details with elaboration in your query, am sure others can chip in as well along with me. Good luck!

[biostart]

The histone modifications at which I am looking are like H3K9me3. They are well known for marking large heterochromatin regions (they also form narrow peaks but I am not interested in these; I am specifically interested only in broad peaks).

Here is the command line with MACS:

macs14 --format BED -t filename1.bed -c filename2.bed -n filename3

Here is the command with HOMER:

findPeaks DirName -style histone -o auto -i DirName2

[ivivek_ngs]

You are using the wrong MACS version. Macs14 was not designed to find broad peaks. Check for MAC2 in github you will find the new macs2 that is well equipped with broad calling. MACS2

Answer 1

[Petr Ponomarenko]

Have you tried findHotspots.pl from DiffReps https://github.com/shenlab-sinai/diffreps ? I started using it just recently and would love to know what Biostars community think about it and tools for broad peaks analysis in general as biostart asks.

[biostart]

unfortunately this comment does not address the question of the topic

Answer 2

[Bohdan Khomtchouk]

Give this a shot: http://biorxiv.org/content/early/2017/03/29/121897

[Ian]

Looks interesting; a way of complementing traditional peak calling methods - thanks.

Answer 3

[Rashedul Islam]

Hi, if you expect peaks of 10-100kb width, you can merge nearby narrow peaks using beedtools merge -d option. I found another peak caller FindER that allows controlling the peak width with -minSize (minimum size of peak), -maxGap (maximum size of peak) parameters.