ChIP-seq peak caller that handles biological replicates

[AnnaE]

Hi guys,

I'm quite new to the field and trying to find my way around all available resources is a challenge so I'm asking for some advice here.

I've preformed ChIP-seq for a transcription factor (thus expecting "narrow" peaks) in biological triplicates complete with input controls.

To analyze my data I would like a peak caller that can make use of the biological replicates, is there one out there?

I know I can run each replicate separate and overlap common peaks but if there is software that can make use of the replicates and turn it into statistics I'd prefer that option.

Best regards,

Anna

[dariober]

Hi guys,

I'm quite new to the field and trying to find my way around all available resources is a challenge so I'm asking for some advice here.

I've preformed ChIP-seq for a transcription factor (thus expecting "narrow" peaks) in biological triplicates complete with input controls.

To analyze my data I would like a peak caller that can make use of the biological replicates, is there one out there?

I know I can run each replicate separate and overlap common peaks but if there is software that can make use of the replicates and turn it into statistics I'd prefer that option.

Best regards,

Anna

Hi Anna,

It seems what you are looking at is something like the Bioconductor package DiffBind http://www.bioconductor.org/packages.../DiffBind.html

So you your workflow could be:
- For each library, align reads to your reference genome
- Call peaks separately for each library (using he input control) using your favourite peak caller (e.g. macs)
- Pass aligned reads (bam files) and peak sets to DiffBind for differential binding analysis making use of biological replicates

Hope this is what you are after...

All the best
Dario

[mudshark]

i think CisGenome (www.biostat.jhsph.edu/~hji/cisgenome/) can do exactly what you want.

[AnnaE]

Thanks for taking the time to answer, I appreciate it!

@Mudshark: Thanks, that looks very interesting indeed. I'll be trying it out for sure!

@dariober: At this stage I'm not looking for differential binding, I only have ChIP-seq from one condition so what I want is TF binding map so to speak.

Best regards!
Anna

[jbrwn]

Thanks for taking the time to answer, I appreciate it!

@Mudshark: Thanks, that looks very interesting indeed. I'll be trying it out for sure!

@dariober: At this stage I'm not looking for differential binding, I only have ChIP-seq from one condition so what I want is TF binding map so to speak.

Best regards!
Anna

diffbind was suggested because if you were to use something like macs to call peaks, it would not take biological replicates into consideration. after calling peaks in macs, you would use diffbind to perform the overlapping of the peaks to find peaks called in all biological replicates. similar work could also easily be done using bedtools intersect.

edit: or just use cisgenome as was suggested.

[qtrinh]

Hi,
You can also try IDR:

https://sites.google.com/site/anshulkundaje/projects/idr

Q

[troublezhang]

Hi, for replicated ChIP-Seq data, you could try running PePr (https://ones.ccmb.med.umich.edu/wiki/PePr/).