search

macs3-project / MACS

MACS -- Model-based Analysis of ChIP-Seq
https://macs3-project.github.io/MACS/
BSD 3-Clause "New" or "Revised" License
695 stars 270 forks source link

MACS2 default q-value different in README vs using MACS2 #184

Open rebeccawh opened 7 years ago

rebeccawh commented 7 years ago

MACS2 is wonderful, thank-you for it. I hope it becomes a paper to be referenced at some point.

I've only just noticed that while the README says the default q-vlaue is 0.01, that in the MACS2 (2.1.0) output the default q-value is 0.05. The README indicates that 0.01 is good for strong peaks, while the 0.05 is more for broad/weaker peaks. Yet I found the output of someone asking a question about MACS2 and MACS2's output said the opposite.

qvalue cutoff for narrow/strong regions = 5.00e-02

# qvalue cutoff for broad/weak regions = 1.00e-02

I am now rather unsure whether I have been processing all my DNase-seq and ATAC-seq with the wrong q-value or the right one. The choice of q-value has quite an impact on the regions that are reported.

Also, I was wondering if it was possible to report more decimal places for the q-value or p-value. I've been trying to use IDR with MACS2 and I am wondering if the large number of ties due to only 5 decimal places is causing some problems.

thank-you for any advice on this, Rebecca

IanCodes commented 7 years ago

There is at least one article, but it is behind a paywall: https://www.ncbi.nlm.nih.gov/pubmed/24743991

On 10 April 2017 at 09:27, rebeccawh notifications@github.com wrote:

MACS2 is wonderful, thank-you for it. I hope it becomes a paper to be referenced at some point.

I've only just noticed that while the README says the default q-vlaue is 0.01, that in the MACS2 output the default q-value is 0.05. The README indicates that 0.01 is good for strong peaks, while the 0.05 is more for broad/weaker peaks. Yet I found the output of someone asking a question about MACS2 and MACS2's output said the opposite.

qvalue cutoff for narrow/strong regions = 5.00e-02

qvalue cutoff for broad/weak regions = 1.00e-02

I am now rather unsure whether I have been processing all my DNase-seq and ATAC-seq with the wrong q-value or the right one. The choice of q-value has quite an impact on the regions that are reported.

Also, I was wondering if it was possible to report more decimal places for the q-value or p-value. I've been trying to use IDR with MACS2 and I am wondering if the large number of ties due to only 5 decimal places is causing some problems.

thank-you for any advice on this, Rebecca

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/taoliu/MACS/issues/184, or mute the thread https://github.com/notifications/unsubscribe-auth/AJWJPAkwsOaQl0MjmPiacamzgMFkISUQks5ruefUgaJpZM4M4dzT .

rebeccawh commented 7 years ago

Thank-you but I don't have access to the MACS paper you linked. Also a number of things changed in MACS relative to MACS2 and I believe threshold was one. Additionally, what I wrote above about qval < 0.01 vs. qval < 0.05 is contradictory within MACS2 itself i.e. within the MACS2 README vs the output of MACS2.

taoliu commented 7 years ago

@rebeccawh The default is q=0.05. I haven't carefully edited MACS2 README file. The output from command line is the actual parameter. You can also confirm the parameters from MACS2 running message or the header of xls file.