transcripts missing from cufflinks

[djm2007]

Hi All-

I'm using cufflinks to assemble transcripts with a Drosophila genome annotation file from refSeq (refFlat), cuffcompare to create a "_combined.gtf" file from multiple samples, and cuffdiff to detect differential expression between samples using the "_combined.gtf". I'm particularly interested in the tss_group files since these allow me to look at the behavior of individual transcription start sites between samples.

In general, everything seems to be working properly. My problem is that 10% of transcripts are missing from the tss_group files, despite the fact that these same transcripts have FPKM values, and they are present in the isoform, gene, and other cuffdiff output files.

My initial explanation was that the missing transcripts were removed because they overlap, but I've found many overlapping transcripts present in the tss_group files.

Has anyone encountered this/have any thoughts?

Thanks,

-Dan

[njdickens]

ditto

I've just discussed something similar using cufflinks with a gtf file, there are (seemingly random) entries missing in my output files. I had thought it was a parsing error but they never make the transcripts.expr output.

[njdickens]

A quick update - I updated to Cufflinks version 0.9.3 (the most recent) and the problem is fixed. I still don't know why, etc but moving on. Some of the method also seems improved with the recent version so it is well worth an upgrade.

[dnusol]

Dear njdickens and all,

I am bringing up this old thread because I am facing a similar issue. I have run a tophat-cufflinks-cuffcompare-cuffdiff workflow on a set of 185k assembled contigs by velvet-oases analysis for two samples.

Cufflinks 2.1.1 reports all 185k loci being processed, as follows

Assembling transcripts and estimating abundances.
> Processed 185073 loci.

but then the output is drastically reduced to a final cuffdiff output of:

Processed 28990 loci.
Performed 29220 isoform-level transcription difference tests
Performed 29110 tss-level transcription difference tests
Performed 29074 gene-level transcription difference tests

I am trying to rescue some of the original loci to see how they behave in terms of FPKM, since I know that after tophat alignment, there are numerous reads aligning to them. But I cannot get them back, so they have somehow been discarded at some point during cufflinks analysis step. Cufflinks has a plethora of options. Can anyone tell me how can I tune cufflinks so that I get information to a larger proportion of the original loci?

Cheers,

Dave