I just checked and Cufflinks v0.9.1, which is the version in use on our public server, produces p_id when given the -s option, so it's not needed to address this issue. However, we'll nonetheless try to update to the most recent version of Cufflinks soon.

On Wed, Dec 8, 2010 at 1:24 PM, David Matthews <D.A.Matthews@bristol.ac.uk> wrote:

Hi Jeremy,

Just got this from Cole - the -s option is needed for sure and Adam also thinks that the version needs to be upgraded to the latest one - sorry to add more work to your plate!!

Glad to get to the bottom of it at last - perhaps a post in Seqanswers would be worthwhile...

David

Begin forwarded message:

From: Cole Trapnell <cole@cs.umd.edu>

Date: 8 December 2010 18:19:53 GMT

To: David Matthews <D.A.Matthews@bristol.ac.uk>

Cc: Adam Roberts <adarob@gmail.com>

Subject: Re: Cufflinks

Ah. Yes, we really need to add a note in the manual about this.

Getting the p_ids to work requires both that your GTF file have annotated "CDS" type records (not just "exon" type records like the Cufflinks assembler spits out), and that you also supply a reference genome sequence with -s. David, is that what you're doing?

C