best pipeline for Chip-seq analysis

[priya]

Hi,
I have Chip-seq data, and I would like to identify the binding sites, peak annotations, and then motif analysis. I came across vast tools for each of the analysis from google.
I have tried experimenting some of the tools on my data

Peak Calling - MACS, SISSRS, HOMER-findPeaks

Peak Annotations - HOMER-annotatePeaks ,peakAnalyzer

Motif Analysis - HOMER- findMotifs, MEME-CHIP, OPOSSUM-3


I tried all the above, but I would like to know what others are mostly using and which is the best and highly cited by publications?

Any suggestions please?

[TiborNagy]

Most of the publications use MACS for peak calling. The motif analysis is a bit tricky. As far as I know there is no 'golden rule' for this task.

[A.N.Other]

I should think MACS is the most widespread for detection (excluding those that don't pay for something like Avadis or some other paid-for suite). Then MEME / MEMEchip for the analysis of the regions.

[priya]

But I see that MACS peaks are very broad , in comparison with SISSRS for example.
Is there anything like after identifying the peak sites by MACS,do I need to use any other tool to minimize the peak lengths ?? I feel that broad peaks makes difficulty in identifying the consensus sites..

[A.N.Other]

Well chip peaks can legitimately vary from a few 10s of bases to kilobases and beyond, so it's tricky to say what is 'correct' without knowing more about the dataset. What were you doing the chip for - are you expecting narrow peaks?

[priya]

Well chip peaks can legitimately vary from a few 10s of bases to kilobases and beyond, so it's tricky to say what is 'correct' without knowing more about the dataset. What were you doing the chip for - are you expecting narrow peaks?

I am looking for the transcription factors.. and it is known that most of the motifs for TFS are found mostly around 50-75 bp +/- from the peak center...

[Chipper]

But I see that MACS peaks are very broad , in comparison with SISSRS for example.
Is there anything like after identifying the peak sites by MACS,do I need to use any other tool to minimize the peak lengths ?? I feel that broad peaks makes difficulty in identifying the consensus sites..

Use the summits.

[A.N.Other]

Use the summits.

Not necessarily quite that simple - I've had macs center peaks in very strange places in some of my datasets in the past, and I'd agree that it's also a bit prone to elongating peaks. Depending on what you're wanting, there are better tools out there.

Try using findPeaks. It's very good at finding small 'neat' peaks for the sort of thing you're doing.

[priya]

Use the summits.

I am not aware if this summit .. I see in MACS summit file, the binding sites of one bp . Are summit regions are exact peak centered regions?
I came across from the following link how to extract the +/- regions of summits,

https://groups.google.com/forum/#!topic/macs-announcement/vuvNT1IuiRA

[priya]

Not necessarily quite that simple - I've had macs center peaks in very strange places in some of my datasets in the past, and I'd agree that it's also a bit prone to elongating peaks. Depending on what you're wanting, there are better tools out there.

Try using findPeaks. It's very good at finding small 'neat' peaks for the sort of thing you're doing.

Ok, I check that how it is working with my data, Thank you

[nike00]

Not necessarily quite that simple - I've had macs center peaks in very strange places in some of my datasets in the past, and I'd agree that it's also a bit prone to elongating peaks. Depending on what you're wanting, there are better tools out there.

Try using findPeaks. It's very good at finding small 'neat' peaks for the sort of thing you're doing.

Hi,

I would like to have a feedback about findPeaker of Homer.
I am using it and also MACS on the same dataset and I found much more peaks with Homer than with MACS (~1500 MACS-peaks and ~150,000 HOMER-peaks)

Do you know why?

Maybe I need to put some filter on the Homer results or I need to use some options (and not the default?)

My data concern a transcription factor, so this is the command line I am using:
findPeaks STAT3_NT -style factor -o auto -i INPUT_NT

Do you have any experience about this big difference between these results?

any suggestions, please?