[maker-devel] default gene model by quality_filter.pl is the same with those by setting keep_preds=0?

[Carson Holt]

Any hint based predictions rejected for one of the non-AED filters will still be rejected regardless of keep_preds=1.

However you are right in that any raw predictions that were rejected by other filters may still be kept by using this workflow, since keep_preds=1 essentially rescues all non-overlapping raw predictions that would have been rejected.

—Carson


> On Feb 17, 2017, at 2:57 PM, Quanwei Zhang <qwzhang0601 at gmail.com> wrote:
> 
> Hello:
> 
> I am following the Maker protocol (2014) to rescue rejected gene models. I set keep_preds=1 when I ran Maker, and then used "quality_filter.pl <http://quality_filter.pl/> -d" to get the "default" Maker report. According to the annotation it will "Prints transcripts with an AED <1". 
> 
> But in a previous post (the link below), it was said "there are models with AED less than 1 that get rejected for other reasons. ... For example if the only evidence is a protein alignment that has deep overlapping HSPs (extremely low complexity alignment) it will be filtered out even though AED is not technically equal to 1. Also if the overlapping protein evidence is in a different reading frame than the model it is supposed to support then the AED will be less than 1 but eAED will be 1 (extended AED), and the model will be rejected."
> 
> So I wonder, whether the "default" result obtained by "quality_filter.pl <http://quality_filter.pl/>" are really the same as those achieved by setting  keep_preds=0? Will the script also consider other reasons besides "AED <1"?
> 
> https://groups.google.com/forum/#!searchin/maker-devel/The$20reason$20there$20are$20differences$20between$20the$20runs$20is$20that$20there$20are$20models$20with$20AED$20less$20%7Csort:relevance/maker-devel/97aNJkT3bgk/W0MEyqEZAAAJ <https://groups.google.com/forum/#!searchin/maker-devel/The$20reason$20there$20are$20differences$20between$20the$20runs$20is$20that$20there$20are$20models$20with$20AED$20less$20%7Csort:relevance/maker-devel/97aNJkT3bgk/W0MEyqEZAAAJ> 
> 
> Thanks
> 
> Best
> Quanwei
> _______________________________________________
> maker-devel mailing list
> maker-devel at box290.bluehost.com
> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://yandell-lab.org/pipermail/maker-devel_yandell-lab.org/attachments/20170220/fbb4f109/attachment-0003.html>