isoforms.fpkm_tracking also contains genes?

[kitinje]

Hello,

I can't solve a problem with my cufflinks/cuffnorm output.
I'm using gencode v19 GTF file. Read were aligned to genome using STAR.

I have problems with the output in isoforms.fpkm_tracking file.
While I have a correct genes.fpkm_tracking output where only the genes are listed, in isoforms.fpkm_tracking I have both the transcripts (ENST*) AND the genes (ENSG*) listed and quantified.
But fpkm values of the genes are not the sum of the single isoforms, and in some cases transcripts have higher fpkm values then the associated genes.

Why do you think this happens?
Thank you
Marco

[lch]

Hi Marco,

I encountered the same problem as you when I was trying to use cuffnorm on gencode v19 gtf file. There could be two issues.

1. There are some additional fields that are required by cuffdiff/cuffnorm. They can be added using cuffcompare according to the cufflinks manual page.

2. Somehow cuffnorm does not recognize the entries for genes in the gtf file (3rd field as "gene"). It will treat these entries as transcripts. Perhaps you can remove them first.

Hope this help.