Can F-seq appers the p-value for each peak in the output?

Question

[vw]

Hi all,

I am using F-seq to call peaks for my ATAC-seq data. But I found it do not provide P-values for each peak. Is there any way to generate P-values from F-seq tool?

thanks a lot! Yifei

[ATpoint]

Never used it but according to the documentation it outputs narrowPeak, so $8 would be the nominal p and $9 the FDR-adjusted p.

[vw]

Hi ATpoint, I just tried to set the output as narrowPeak. The head of the output file is:

chr10   3182358 3182372 chr10.1 0       .       0.01970927      -1      -1      0
chr10   3332962 3333155 chr10.2 0       .       0.03675616      -1      -1      135
chr10   3365943 3366844 chr10.3 0       .       0.08080269      -1      -1      564

Negative one was assigned to column 8 and 9. What's that meaning?

[ATpoint]

I do not know as I have no experience with F-seq. I like using MACS2 as it is an established and most importantly well-maintained tool. For this, I write the ATAC-seq cutting sites (= the 5' ends of both the paired-end reads) to BED format (bedtools bamtobed) and then use:

macs2 callpeak -t cutsites.bed.gz -f BED --nomodel --extsize 100 --shift -50

Works well for me. I prefer to have the q-value at 0.01 so 1% false-discovery rate to get rid of peaks with low counts. If there are replicates, I use the IDR procedure (irreproducible discovery rate) to keep high-confidence peaks.

[vw]

Thank you ATpoint!

I consider deeper for my project. I found the F-seq filters the peaks by a background model rather than q values. Thus, it cannot generate q value. And my project doesn't really need the q value.

But your answer is very helpful. Thank you again!