the gene_id column is different for each sample when counting fpkm in galaxy cufflinks

Question

[zizigolu]

Hi,

I used galaxy cufflinks to count fpkm in my accepted_hits.bam but I saw the gene_id columns entities is not the same for all of the samples. Then after extracting fpkm column for each sample, I should use which gene_id column because there is not any union gene_id column between samples. Do you have any ideas please?

Thank you

[Devon Ryan]

Did you run cuffmerge and then recalculate things? If not, you're comparing apples and oranges.

[zizigolu]

thank you Devon, after long time someone replied my question in biostars!

yes I run cuffmerge by cuffmerge -g genes.gtf -s genome.fa -p 8 assemblies.txt and I have a merged_gtf now but galaxy cufflinks only need bam file and gtf then how I can figure out the connection between the result of cuffmerge and counting fpkm in galaxy cufflinks??? I used the merged_gtf resulted from cuffmerge as reference in galaxy cufflinks but in gene id column I don have AGI id anymore and instead I have XLOC_000001

then what is the solution please?

[Devon Ryan]

Is cuffquant in Galaxy? I don't think we have that on our local install, but perhaps the one you're using has it. Ideally, you would run cuffmerge, which produces a merged GTF file, and then cuffquant with the resulting GTF file and the initial BAM files. If cuffquant isn't available, you can probably run cuffdiff and still get the RPKMs. Yes, it seems a bit silly to run cuffdiff just for the RPKMs, but if cuffquant isn't around that's likely the fastest route.

BTW, things like XLOC_0000001 are created for novel genes or isoforms, so it's often the case that they'll never have a more meaningful ID unless you do some manual curation.

[zizigolu]

thank you,

cuffquant result names Galaxy63-[Cuffquant_on_data_50_and_data_41__Abundances.cxb].bin and unreadable by kate

[Devon Ryan]

cuffnorm should be able to take those cxb files and produce FPKMs.