HIDATA in Cuffdiff

[aquleaf]

Hi all. Sorry to bother you.

I have a question about Cufflinks suite. Basically, I used Cufflinks to assemble transcripts and genes, then Cuffmerge to merge all the assemblies and generate an annotation and at last Cuffdiff to identify differentially expressed genes. Some genes were marked with HIDATA when using default value of --max-bundle-frags. After increasing the --max-bundle-frags value, there is no gene labled with HIDATA in the gene and transcript assembly step by Cufflinks. However, in Cuffdiff step, I still got HIDATA tag.

Is there any other arguments should be defined in Cuffdiff step?

Wish your help! Thanks very much! I really appreciate it.

[sdriscoll]

did you try setting --max-bundle-frags when you ran cuffdiff?

[aquleaf]

did you try setting --max-bundle-frags when you ran cuffdiff?

Yes, I increase --max-bundle-frags in cuffdiff step, but still got HIDATA

[sdriscoll]

I can't say for sure but maybe this this is happening in cuffdiff and not Cufflinks because the first thing cuffdiff does is pool all of the alignments from your replicates and quantifies it all at once. They do that to establish a confidence interval against which they can compare replicate expressions to pick out isoforms with outlier expressions so it can skip them in testing. You might just need to raise the --max-bundle-frags setting by a factor equal to the number of samples you're running. Or set it to 999999999. That should be well beyond any single locus' depth.