Hi,

I have been following the csaw vignette and this article: https://f1000researchdata.s3.amazonaws.com/manuscripts/8237/a7de2136-b076-43a6-9376-7eec24c7d4117016-_Aaron_Lun_V2.pdf?doi=10.12688/f1000research.7016.2

For estimating the fragment length, two commands are suggested: maximizeCcf (in the vignette) and which.max (in the article). However these commands are giving two different results:

`> which.max(fragmentLength) - 1

[1] 40

> maximizeCcf(fragmentLength)

[1] 192`

I guess that the correct way is the "maximizeCcf(fragmentLength)" command but why using "which.max(fragmentLength) - 1" in the article?

Thanks.

I added maximizeCcf after the article was published, hence the difference between the two documents. It's a bit of a pain to update the F1000 article (as I have to go through proofing, etc.); but I do keep the Bioconductor version of the workflow up to date:

http://bioconductor.org/help/workflows/chipseqDB/

The reason why your values are different is probably because of a "phantom peak" at the read length. maximizeCcf skips over maxima at short delay distances and tries to report the local maxima corresponding to the fragment length, which is what we really want.