Annotating 3' UTR in the central nervous system

[ccros]

Hi,
I am interested in identifying 5' UTR and 3' UTR of known mouse genes linked to olfaction. I have access to the CDS, poor annotations of the 5' and 3' from Biomart, CAGE data for the promoters and lots of RNASeq data (10 samples with 30-80M reads, paired-end but unstranded) from the litterature.

I have read (Widespread and extensive lengthening of 3' UTRs in the mammalian brain Miura 2013) that in the CNS there are longer 3' isoforms, often tied to local transport and translation in dendrites or axons.
Do you know what type of tissue is used for 'official' annotations (mm10 for example) done with RNA Seq data? In IGV I see very different types of 3' UTR in the olfactory epithelium, but this could be a tissue specific thing.

What would you use for annotating 3'UTR?
I am thinking of trying out IsoSCM (Jan2015, Shenker), but is very recent. I have also heard some good about TrinityRNASeq. Cufflinks is rather bad for this task, it missed isoforms I could see on IGV (and then confirmed with Northern Blots).

Thanks a lot for your help.
Cyril

[dpryan]

The official annotations are typically from a mix of different tissues, since they end up being drawn from databases of observed transcripts. In my mouse brain RNAseq samples, I almost always see longer 3'UTRs than exist in the Ensembl/UCSC/etc. annotations. I have primarily hippocampal samples, so I wouldn't expect much coverage of the receptors you're interested (so I can't offer much insight into them).

One common method if you absolutely need the UTRs is to simply run cufflinks on the samples followed by cuffmerge with the reference annotation. I've tried that in the past and it seems to work fine (not that it changed the results of the down stream statistics much, though perhaps it will for you).

[ccros]

Thanks for the answer.
So far, IsoSCm has given me greatly improved 3' UTRs. There are quite a few options to set. For example, short introns (<100) seems to be ignored in 3' UTR.
I also note that new software (early 2015) such as Kleat and other de novo transcriptome assembly tools offer the same features.