[maker-devel] some strange examples of maker annotation

[Carson Holt]

maker seems to prefer to select the snap gene mode, but not genemark

> Genemark generally scores lower, and has a very large tendancy to overlap
> transposons (it can't handled masked fragments, so has to be run on the
> unmasked genome).  Looking through the code base, I now see a section where
> the non-overlapping model is set to always exclude genemark from the
> non-overlapping consensus set if there are masked gene predictors such as snap
> or augustus, and to only accept it's models when the evidence supports it.
> I'd need to filter genemark candidates for transposon overlap before I could
> lift this limitation.
> 

Fig 1. the snap gene mode of non_overlapping_ab_initio is redundant
(overlapping) with the maker gene annotation.

> The non-overlapping is stranded.  These are on different strands. This really
> does happens in eukaryotes, so if the evidence supports it we have to allow
> it, and if you set keep_preds=1 you can get it just because the gene predictor
> supports it reguardless of physical evidence support.
> 

Fig 2. the snap gene mode of non_overlapping_ab_initio is redundant
(overlapping) with the maker gene annotation.
> 
> On different strands.


Fig 3. there is gene redundancy even within the maker gene annotation

> They are on opposite strands.
> 

Fig.4 no evidence support the snap gene mode. augustus and genemark have
similar results but different from snap. But the snap gene was selected as
non_overlapping_ab_initio

> Try using Apollo rather than IGV, it becomes so much more obvious because
> apollo separates the strands into separate panels.


Thanks,
Carson



From:  刘慧泉 <liuhuiquan at nwsuaf.edu.cn>
Date:  Thursday, 18 April, 2013 9:37 AM
To:  Carson Holt <carsonhh at gmail.com>, <maker-devel at yandell-lab.org>
Subject:  some strange examples of maker annotation

Hi Carson,
I run maker on my  genome with “keep_preds=1” or “keep_preds=0”
respectively. When I manually check the results of maker in Integrative
Genomics Viewer (IGV), I found most of the genes annotated by maker were
good. But I also view some strange examples for the results. I don’t know
how to inteprete these. hope you can give me some suggestions. please see
the attached file.  thank you very  much.
best regards,
Huiquan


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://yandell-lab.org/pipermail/maker-devel_yandell-lab.org/attachments/20130418/336920d8/attachment-0003.html>