Why cufflinks package merge more than one gene name in the same transcript?

Question

[rodolpho.gheleri]

Hello,

I am running an experiment with 4 samples among 2 conditions (Control vs Mutation) and 2 replicates of each one.

I am mapping the reads with segemehl and running cufflinks package to analyse differential expression in HOX genes.

Looking at the gene_exp.diff from cuffdiff output, I found entries like this below. Why this happens?

XLOC_019730     XLOC_019730     HOXC10,HOXC4,HOXC5,HOXC6,HOXC9,RP11-834C11.12   chr12:53981508-54056030 C       MUT     OK      26.9654 36.546  0.438607        0.940684        0.18165 0.997078        no 

My steps/commands done in this experiment can be found here.

Answer 1

[colin.kern]

This happens when cufflinks has built a transcript that spans multiple genes. Did you do any trimming of your reads before aligning? That might improve your results. Also if your library is stranded, you will need to tell cufflinks that (look at the --library-type argument).

[rodolpho.gheleri]

Thanks for the answer. I am not trimming my reads before aligning, just after that. I am using RseqC split_bam.py) to remove rRNA from mapped reads by segemehl. Can you recomend some approach?

Answer 2

[andrew.j.skelton73]

The gene level analysis in the tuxedo package works by selecting a 'window', which is a locus in which all reads in that region are summarised to give a gene level quantification. So your XLOC code, is a particular region rather than a gene, that XLOC happens to encapsulate all of those genes you've given. I've never been able to figure out the way that Tuxedo specifically derives an XLOC, but it does have something to do with the way the ends of transcripts overlap. I'd recommend you look at the htSeq -> DESeq2 route for differential gene expression analysis.