Hi!
Is there a way to rank the peaks in broad peak file of macs2?
The fifth column of the file contains value 100*-log10pvalue, can this be used to rank the peaks? or can this column be converted into -log10pvalue by dividing each value in column by 100?
I know it is a very naive question, but I just need to be sure what I am doing is right.
Thank you
From what you describe that multiplier is there only to produce values as percent. You don't need to do anything with it just use it directly.
That being said ranking anything by p-value is usually a bad idea (although using them as such is very common). A p-value is the likelihood of the event being observed by random chance, but it is not indicative of biological significance, correlation to an outcome or general importance.
Use the p-value to separate peaks into those that you trusts and those that you don't trust. Then perform a different analysis on them.
If you want the p-value, you should divide by 100. What you do with the p-values, once you have them, is up to you. I agree with Istvan that the p-value is not a perfect way to quantify the biological significance of a peak. (A p-value is actually how extreme your peak is in the context of the null hypothesis. So the usefulness of the p-value is related to how sensibly MACS2's null hypothesis is in your application). However, you will often see people using it this way. The p-value comes from the how extreme the tag counts are relative to the background. So, quite often big peaks are more significant. It's an interesting question whether you can relate the size of the peak to the strength of binding of your factor. I don't think this has been resolved in the field.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
The author Tao Liu has a very active google group. You might want to ask him directly. The link to join is here: http://liulab.dfci.harvard.edu/MACS/Download.html He typically answers questions very quickly.