How is "integer score for display" calculated in MACS2

Question

[Vanilla]

Hi all,

In the MACS2 output file "NAME_peaks.narrowPeak", there's a column called "integer score for display". I found the score is not proportional to the 7th column "fold-change".

For example, I have two peaks. The "integer score for display" and "fold-change" for these two are:

peak1 207 3.21293

peak2 268 3.16602

So for peak1, the "fold-change" is higher than peak2, however, its "integer score for display" is lower than peak2.

And the "integer score for display" is also not proportional to "pileup height at peak summit" (6th column in "NAME_peaks.xls").

As I didn't find any description about this score in their papers, I'm wondering how the score is calculate? How could I understand this "integer score for display" properly? I mean how this score represent the intensity/affinity of TF binding?

Thanks all!

Answer 1

[igor]

This was actually addressed by the MACS developer:

The score is for 'display' purpose only. It's for UCSC browser and will make stronger peaks darker, and weaker peaks lighter.

Source: https://github.com/taoliu/MACS/issues/84

[Vanilla]

Hi Igor,

Thanks very much for your answer. Actually their explanation is still vague, and I still didn't get how this score can reflect how strong the peak is.

But anyway, as they suggested using column7-9 for downstream, I may only take this score for visualization but not used for filtering.

Thanks again for your help.

[SplitInf]

since this is one of the top search results, I'm cross-posting this comment from another user (alexbarrera)

"that the values of the 5th column in narrowPeak files are equal to the integer part of 9th column (-log10qvalue) multiplied by 10. int(-10*log10qvalue)" https://github.com/macs3-project/MACS/issues/123

Note that as stated in the link, in some cases pvalue is used for the conversion instead.

Answer 2

[g.guiducci]

I agree with Vanilla, I don't get how the score is calculated and I am trying to answer to the same question.