Page 1 of 1

How do you know that....

Posted: Wed Jan 09, 2013 10:57 pm
by cbazinet
In annotation, there are occasionally small exons predicted by multiple gene prediction programs, with an RNAseq signal, but no significant BLAST hits to other Drosophila. What do you make of these? e.g., if biarmipes is expressing a small genes that is unique to it, that's what we would expect, and that could be the basis of some biarmipes-unique biology. But the lack of a BLAST hit argues for ignoring it. Nonetheless, occasionally there will be a gene unique to the species we are looking at, so ignoring these potentially walls us off from learning about some potentially unique genes?

Re: How do you know that....

Posted: Thu Jan 10, 2013 11:00 pm
by wleung
You are correct that the GEP annotation strategy would not allow you to annotate species-specific gene (e.g. genes that are only found in D. biarmipes). The problem is that we cannot confidently construct the gene models using only RNA-seq data. For example, while the oases and cufflinks transcripts tries to predict different isoforms, in general they are not very accurate because they can only resolve overlapping exons in the different isoforms if there are spliced reads and that expression levels of the different isoforms are not the same.

Consequently, while we can construct a predicted gene model based on the gene predictions and the RNA-seq data, we cannot confirm the gene model without additional experimental data or conservation in the nearby Drosophila species.