Cufflinks p_id explanation

[fongchun]

Hi,

I've been using cufflinks/tophats to work with 28 RNASeq libraries (some of which below to group A, and some to group B) and so far I've the following workflow:

1. Ran tophat on each library with refseq gene model annotations
2. Ran cufflinks on each library with refseq to Reference Based Assembly
3. Ran cuffmerge on all the libraries with refseq
4. Running cuffdiff between two groups merged.gtf from cuffmerge

When I look at the merged.gtf file produced from cuffmerge, I see the attribute p_id added to some of the lines. According to the cufflinks website, p_id means:

The ID of the coding sequence this transcript contains. This attribute is attached by Cuffcompare to the .combined.gtf records only when it is run with a reference annotation that include CDS records. Further, differential CDS analysis is only performed when all isoforms of a gene have p_id attributes, because neither Cufflinks nor Cuffcompare attempt to assign an open reading frame to transcripts.

I've been struggling to understand exactly what this even means. They must be referring to the CDS features in my refseq.gtf file? Cuffdiff is supposed to output a cds_exp.diff file which according to the website:

Coding sequence differential FPKM. Tests differences in the summed FPKM of transcripts sharing each p_id independent of tss_id

When I look at the merged.gtf file, I have yet to see transcript share the same p_id. Only exons of the same transcript share the same p_id. Am I missing something here? What does the p_id attribute even truly mean? Is my pipeline wrong here or something?

Thanks,

Fong

[fongchun]

Does anyone have an answer to this question? Or some more reference as to where I can find this information?