Gene expression of duplicate genes

[krespim]

I have some polyA(minus) RNAseq data and would like to quantify the gene expression of Pol3 transcribed genes. Well, quite a lot of these are duplicate loci (such as the RNU6) which makes it very difficult.

I have used cufflinks but results do not appear to reflect what it is know about the expression of some genes. I suspect that because of the multi-mapping issue gene expression is being underestimated. I was thinking about having a got at MMSEQ. Can anyone recommend a good strategy for this? Someone mentioned to me that extending the gene region to count those reads that extend beyond the gene body. Would this be a good idea?

[Richard Finney]

Mis-mapping because of sequence similarity to many regions is indeed a problem.
Retrogenes, cryptic duplicates, other homology can make analysis difficult.

You are right to avoid "cargo cult" science. Just because you ran cufflinks, doesn't mean it's producing truth. Just because you setup a bamboo runway and communication huts doesn't mean valuable cargo will show up from the sky.

Techniques I use is to keep a list of known or suspected duplicate regions AND ALSO filter based on "self chaining". Just throw them out as unreliable. The Self Chain track is available from UCSC.