DiffBind returns empty consensus peakset
Entering edit mode
Last seen 4.9 years ago

Dear Rory,

I read peaksets to DiffBind as follows:

wt <- dba.peakset(NULL, peaks = system.file("/home/labs/shlush/avivdm/chipseq31375/9macsbroad/wt1ipchip7-5.NOMODpeaks.broadPeak", package = "DiffBind"), peak.caller = "narrow", sampID="WT1_75", tissue = "K562", factor = "H3K9Me3", condition = "75", replicate = 1)

wt <- dba.peakset(wt, peaks = system.file("/home/labs/shlush/avivdm/chipseq31375/9macsbroad/wt2ipchip7-5.NOMODpeaks.broadPeak", package = "DiffBind"), peak.caller = "narrow", sampID="WT2_75", tissue = "K562", factor = "H3K9Me3", condition = "75", replicate = 2)

wt <- dba.peakset(wt, peaks = system.file("/home/labs/shlush/avivdm/chipseq31375/9macsbroad/wt1ipchip31-3.NOMODpeaks.broadPeak", package = "DiffBind"), peak.caller = "narrow", sampID="WT1_313", tissue = "K562", factor = "H3K9Me3", condition = "313", replicate = 3)

wt <- dba.peakset(wt, peaks = system.file("/home/labs/shlush/avivdm/chipseq31375/9macsbroad/wt2ipchip31-3.NOMODpeaks.broadPeak", package = "DiffBind"), peak.caller = "narrow", sampID="WT2_313", tissue = "K562", factor = "H3K9Me3", condition = "313", replicate = 4)

and then:

wt_consensus <- dba.peakset(wt, 1:4, minOverlap = 4, bRetrieve = TRUE)




4 Samples, 0 sites in matrix:

   ID Tissue  Factor Condition Replicate Caller Intervals

1 WT1_75 K562 H3K9Me3 75 1 narrow 0

2 WT2_75 K562 H3K9Me3 75 2 narrow 0

3 WT1_313 K562 H3K9Me3 313 3 narrow 0

4 WT2_313 K562 H3K9Me3 313 4 narrow 0

Since all the previous stages of the analysis (according to your tutorial) were OK, this baffles me. What is wrong in what I am doing? Sincerely, Aviv.

DiffBind consensus • 1.0k views
Entering edit mode
Rory Stark ★ 5.1k
Last seen 22 hours ago
Cambridge, UK

This is showing that zero sites have been loaded.

One unusual thing I see is that your use of system.file() to specify the peaks directories. As you are giving full paths to the peaks (starting with "/home"), you shouldn't need to use system.file() and specify them as being installed with the DiffBind package. So do try this again with just the path names, and not the system.file() call (or package="DiffBind").

I just checked and I see that if you load a peakset using dba.peakset(), and the file is not found, it adds a peakset with zero peaks instead of reporting an error (file not found). So the most likely answer is that the file paths you are giving are invalid. For example, the first file is "/home/labs/shlush/avivdm/chipseq31375/9macsbroad/wt1ipchip7-5.NOMODpeaks.broadPeak", but I see some italics in there -- are there some characters that have been altered? You can do a file.info() call with the path you are passing in to see if the file is acessible form R and contaions data.

Finally, although in this case the issue is that no peaks are being read in, if you specify minOverlap=4 for a dataset with four peaksets, you are not guaranteed to have any consensus peaks -- you will only get something if there are overlapping intervals in all four peaksets.


Entering edit mode

Thanks, Rory, for your prompt answer. What solved it was passing the path w/o system.file() and w/o package="DiffBind". And I do have peaks when I set minOverlap=4, which I have to look deeper into. Thanks again.


Login before adding your answer.

Traffic: 565 users visited in the last hour
Help About
Access RSS

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6