Does the countToFPKM package work for single end RNA seq or its only for paired end?

Question

[salehm]

Hi All, I have a featurecount output row counts and need to convert them to FPKM. I knew about the countToFPKM package, however, the description of the package says its for paired end RNA seq data. My data are single end. Could I still use this package? and if not could you please suggest an easy way to do it?

The reason I need to convert from row counts to FPKM is to perform ssGSEA analyses on gene pattern. Thanks.

Mohamed

[Ram]

Does the FPKM metric apply to SE data? What would be the difference between RPKM and FPKM in the context of SE data?

[salehm]

It sounds that yes the FPKM needs paired end. So I may need to convert my single end data to RPKM, any easy idea how I can do that? Thanks.

[Ram]

What have you tried? Have you searched online for "get RPKM from raw counts"?

[salehm]

I was thinking of using countToFPKM package but it says it works for paired end reads, I am not sure if I could also use it for single end reads.

[Ram]

So you searched for "get RPKM from counts" and the only thing you see is the countToFPKM package?

[swbarnes2]

Is featurecounts going to do FPKM right? Don't you need something like Kallisto or RSEM that understands how the different transcripts are proportioned?

Answer 1

[seidel]

Once you have a matrix of counts per feature, there is no way to tell if the counts for the feature came from paired-end data or single end data. The single or double endedness of your DNA Fragments is accounted for in the generation of an integer count representing a feature. Thus, your count matrix can be converted to FPKM whether it was generated from single or paired-end data. FPKM is used to represent abundance regardless of end type, and is identical to RPKM for this same reason. RPKM simply suffers from a historical prematurity in that "Reads" actually represent "Fragments", whether sequenced by one end or two. One caveat: the package, and the method it implements (from Trapnell et al., 2010), uses the average DNA fragment length for a given library in the calculation. This can be determined easily in paired end data, but for single end data you usually have to estimate the library insert size (usually an educated guess).

[salehm]

Thank you so much for your clear and informative answer @seidel