Improve transcriptome by merging transcripts putatively from the same locus

[mchapman47]

Sorry for the long Title!
We have a de novo assembled transcriptome and have become aware that a portion of the transcripts are in fact different portions of the same locus (sometimes with a short region of overlap but often not). Im sure this is common but its rarely commented on. I was wondering if there was a way to use a distant (several millions of years) but very well annotated transcriptome (e.g. Arabidopsis) to pair up transcripts that seem to come from the same locus. One concern is that gene duplication etc might mean that the results arent very reliable... but Id like to give it a try!
Thanks a lot!
Mark

[Jeremy]

Hi, I found a similar thing with a de-novo assembly of 454 transcriptome data. The problem is poor coverage resulting in an insufficient overlap or no overlap between regions. I also tried to group reads by comparing to Arabidopsis to overcome this, but in my case it ended up grouping way too many reads that were obviously different genes so I decided to just leave it as it was and accept that some reads are in fact two portions of the same gene. My problem was that the species I was working with was too different and more complex than Arabidopsis. Depending on the species you use you may have better luck than me.