Cuffdiff missing genes/transcripts?

[swarbre]

So i completed a tophat-cufflinks-cuffcompare-cuffdiff workflow using the cuffcompare combined.gtf file as the reference in cuffdiff. My assumption was that the number of genes and transcripts in the the final cuffdiff files e.g. gene_exp.diff should be the same as in the input Cuffcompare.combined.gtf.

The input Cuffcompare.combined.gtf has 15126 distinct gene_ids while the cuffdiff output file gene_exp.diff has only 9029.

So 6027 genes are "missing"

The "missing genes are all single exon and are either class_code "." OR class_code "u" however not all genes of these classes are found in the "missing" list.

Does any one know if there is a known reason why not all genes would be present in the cuffdiff output files?

[swarbre]

Any ideas????

[Krish_143]

Hello,

Did you found answer !!

[BethSutton]

I am having the same problem. I have one (single exon, class code u) gene missing from my Cuffdiff output that is present in my gtf file, which I made with Cufflinks and Cuffmerge. I also have >100 transcripts (but not their corresponding genes) missing from the Cuffdiff output.

Did anyone find an answer?

[Zapages]

I would adjust adjusting the False Discovery Rate a bit, but more importantly increase the maximum number of fragments allows more gene Ids to be found. I believe the default is 1 Million.

I had to play around with this Maximum Number of Fragments value to something like 10 to 50 Million to get the genes that I was interested to appear.

I hope this helps a bit.