about MACS

[gigigou]

I am using MACS to find peaks of chip seq data.
I get a question about the P value.
General MACS gets many peaks from a sample, much more than SISSRs. But not all the peaks are reliable. So I want to filter the peaks MACS found.
In the command line there is an option -p followed by a p value. If it is set too strictly, e.g. 1e-50, there may be no peaks found, otherwise, e.g. 1e-10, there may be too many peaks.
So I want to ask generally which P value is suitable, i.e., on which P value the peaks found are reliable?
Thank you!

[ishmael]

I would surggest you generate the tdf file of the coverage of genome, and load the tdf and peak lists of different p cutoff of MACS and SISSRs.
And I think the shapes of the peaks of coverage will give you more clues.

[gigigou]

I would surggest you generate the tdf file of the coverage of genome, and load the tdf and peak lists of different p cutoff of MACS and SISSRs.
And I think the shapes of the peaks of coverage will give you more clues.

Thank you.
I'm new in chip seq. And I don't quite understand what is "tdf file of the coverage of genome". Could you please tell me how to generate this file?

[ishmael]

Sorry I didn't explain clearly. tdf is the coverage file format of IGV, a genome browser.
I think you'd better to generate the coverage files of chip-seq, and load them into the browser. And than you may load the peak lists you got from different p-value cutoffs and different softwares. Your eyes will give you some clues which peak list is the better choice from the shape of the peaks.

[gigigou]

Sorry I didn't explain clearly. tdf is the coverage file format of IGV, a genome browser.
I think you'd better to generate the coverage files of chip-seq, and load them into the browser. And than you may load the peak lists you got from different p-value cutoffs and different softwares. Your eyes will give you some clues which peak list is the better choice from the shape of the peaks.

I see.
But what I want is to have a general p value to ensure that I can indeed get some peaks which are reliable. Cause this is automatically done by my pipeline.
According to your words I think there may be no such fixed p value, so I think I'd better make the p value as a parameter for users to determine.
Thank you very much for your patient explaine!

[ishmael]

The quality and nature of different chip are different, I think.
So it is alway not bad to check the coverage manually.
If you want a reliable peak list, think about IDR or MAnorm if the replicate samples are availble.

PS, check your private message box

I see.
But what I want is to have a general p value to ensure that I can indeed get some peaks which are reliable. Cause this is automatically done by my pipeline.
According to your words I think there may be no such fixed p value, so I think I'd better make the p value as a parameter for users to determine.
Thank you very much for your patient explaine!

[DNASpeaks]

Hi Everyone
I am doing chip seq analysis using galaxy. I have paired end data and I am having trouble running MACS on my bam file -
Looks like galaxy takes only eland multi format as input and I am not sure of the format -

Do you think converting my bam files to bed files (by creating a BEDPE format) would help? I do not have bed tools installed, I also would like to know if the BAM files need to be sorted before I convert them into BED files -
I appreciate any help or suggestions.
Thanks.