[maker-devel] Serious Start Codon Problem

[Carson Holt]

This issue just came up this week on another e-mail as well.

One way to fix it, is to edit the Bio::Tools::CodonTable module from
BioPerl (use maker --debug to have maker print out the locations of all
modules it uses).  Such a hack should only be done as a temporary work
around.

You change line 256 from -->
---M---------------M---------------M----------------------------

To-->
-----------------------------------M----------------------------


Maker uses the BioPerl is_start_codon function to determine if the codon
used in a result it receives is acceptable or if it should look upstream
or downstream for a different one. Apparently there are actually 3
acceptable start codons in the standard codon table (2 rare ones and one
common), so making the edit removes the two rare ones from the list.

I'm also preparing a 2.30 release that exports a "strict" codon table to
the BioPerl module, that will be used by default and should fix the issue.

Thanks,
Carson



On 10/12/13 11:06 AM, "Felix Bemm" <felix.bemm at uni-wuerzburg.de> wrote:

>Hi,
>
>I have a serious problems with maker annotations especially the start
>codon placement. It started happening with maker version 2.27! About 1/3
>of my genes are missing a proper start codon. When reviewing the GFF
>files gene predictors and est evidence standalone would predict the
>right gene structure including the start codon. I was thinking that this
>problem was somehow linked to my gene models but looking at their
>standalone prediction I discarded that theory. Just some stats about
>proteins with proper start codon (first column) and missing start codon
>(second column).
>
>Maker		9268	4215
>SNAP		16577	896
>Genemark	18764	290
>
>Numbers look the same when Augustus is used (currently retraining).
>Manually using EST's in Apollo often fixes the problems but I don't want
>to check > 4000 genes for this.
>
>Do you have any idea what could cause such a problem? If you need some
>example gff files I can send you.
>
>Best regards
>Felix
>
>-- 
>Felix Bemm
>Department of Bioinformatics
>University of Würzburg, Germany
>Tel: +49 931 - 31 83696
>Fax: +49 931 - 31 84552
>felix.bemm at uni-wuerzburg.de
>
>_______________________________________________
>maker-devel mailing list
>maker-devel at box290.bluehost.com
>http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org