Is anyone having trouble with tophat's SAM file -> cufflinks?

[peterlchang]

I am having trouble running cufflinks on my own SAM file generated from tophat. I used many SAM files generated from real data as well as simulated data, but I can't get cufflinks to assemble the transcripts using any of my SAM files.

I tried downloading the file: test_data.sam and it works fine. It identifies three exons and notes that they are part of one transcript. When I try this on my SAM file, it identifies all the exons as their own transcripts, even though I see many reads that span the exons.

Any help would be appreciated. Thanks

[Siva]

Did you run Tophat on more than one sequence reads at the same time? If so you might want to run it one at a time. If not, then well, at the moment I can't think of an answer.

Siva

[peterlchang]

Would this matter if I specify a different ouuput file for each run?

I am trying to figure out if the SAM format is a problem or if the actual reads do not cover the junction enough for the transcript to be called.

[peterlchang]

Actually it is really strange. I cannot get any of my tophat-generated SAM outputs to successfully run using cufflinks (it runs but it lists each exon as its own transcript). I've done this with real data and simulated data.

[Siva]

Actually it is really strange. I cannot get any of my tophat-generated SAM outputs to successfully run using cufflinks (it runs but it lists each exon as its own transcript). I've done this with real data and simulated data.

Hi are u working with an organism whose typical intron size is very large? If so you can change Tophat's default parameters. Cufflinks typically gives you a GTF file detailing FPKM etc, a comma delimited transcript.expr file and a list of gene coordinates (genes.expr).

[peterlchang]

The gene I am testing is 34K long (ie. the reference sequence is 34K long). The default intron length for tophat is 500K and I changed that parameter to 50K. The default intron length for tophat is 300K, and that was left unchanged.

In addition, tophat is able to find reads that span the junction. I don't know what the issue could be. I already emailed Cole so I hope he can get back to me asap.

Thanks