[maker-devel] antisense RNA in training set?

[Carson Holt]

MAKER does not require input evidence to be on the correct strand because it performs splice aware alignments via Exonerate against both strands (reverse transcription for the second alignment happens internally). Exonerate should always map spliced alignments to the right strand because it is not be possible to get correct splicing on the opposite strand (splice sites are a stranded feature). The only alignments that are ambiguous are single exon alignments. They are ignored by default, but when not ignored they are stranded to the sequence with the longest canonical ORF.

—Carson



> On Sep 6, 2016, at 1:37 PM, Steven Sullivan <sullis02 at nyu.edu> wrote:
> 
> I have a set of assembled transcripts from a stranded RNA seq run that I want to use for gene finder training in a MAKER run on 'new' organism.
> 
> I've noticed though that some of my assembled transcripts actually appear to be antisense RNAs.    
> 
> Should I include these in the training set?  
> 
> 
> 
> 
> _______________________________________________
> maker-devel mailing list
> maker-devel at box290.bluehost.com
> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org