[maker-devel] Augustus retraining??

[Carson Holt]

Here is some info on training SNAP via the bootstrap technique (i.e. using the models produced by the initial training to seed the next round of training). Even though the examples use SNAP, it would be applicable using the scripts and methods Mikael described in his w-mail —> http://weatherby.genetics.utah.edu/MAKER/wiki/index.php/MAKER_Tutorial_for_GMOD_Online_Training_2014#Training_ab_initio_Gene_Predictors <http://weatherby.genetics.utah.edu/MAKER/wiki/index.php/MAKER_Tutorial_for_GMOD_Online_Training_2014#Training_ab_initio_Gene_Predictors>

Also Jason Stajich wrote an excellent explanation on training Augustus on the GMOD mailing list —> http://brie4.cshl.edu/pipermail/gmod-help/2012-June/001724.html <http://brie4.cshl.edu/pipermail/gmod-help/2012-June/001724.html>
He also included his own scripts to assist with the training —> https://github.com/hyphaltip/genome-scripts/blob/master/gene_prediction/zff2augustus_gbk.pl <https://github.com/hyphaltip/genome-scripts/blob/master/gene_prediction/zff2augustus_gbk.pl>

—Carson


> On Jan 14, 2015, at 3:08 AM, Mikael Brandström Durling <mikael.durling at slu.se> wrote:
> 
> Hi,
> 
> 
>> 14 jan 2015 kl. 09:49 skrev Xabier Vázquez Campos <xvazquezc at gmail.com <mailto:xvazquezc at gmail.com>>:
>> 
>> Hi,
>> 
>> I trained Augustus using the output of CEGMA (http://bioinf.uni-greifswald.de/bioinf/wiki/pmwiki.php?n=Augustus.CEGMATraining <http://bioinf.uni-greifswald.de/bioinf/wiki/pmwiki.php?n=Augustus.CEGMATraining>) through WebAugustus, which makes the training very easy but, and here is my question, can/should I re-train Augustus like it is done with SNAP? And what would I use for the re-training
> 
> I’ve tried an approach of retraining augustus in a manner similar to what has been suggested here earlier for retraining of SNAP. This has been run with a local augustus installation as part of an automated framework I have set up to annotate fungal genomes. Interestingly, augustus seems to converge very quickly. It is not uncommon that autoAugustus reports that it could not improve the initial models that were derived from the CEGMA dataset. Are there other similar experiences on the list? 
> 
> I also a modified version of maker2zff which I call maker2augustus_gff which extracts an evidence set for augustus retraining from the initial round of maker. I’m happy to share it with anyone interested.
> 
> cheers,
> Mikael
> 
> 
>> 
>> Thank you,
>> 
>> Xabier
>> -- 
>> Xabier Vázquez Campos
>> PhD Candidate
>> Water Research Centre
>> School of Civil and Environmental Engineering
>> The University of New South Wales
>> Sydney NSW 2052 AUSTRALIA
>> _______________________________________________
>> maker-devel mailing list
>> maker-devel at box290.bluehost.com <mailto:maker-devel at box290.bluehost.com>
>> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org
> 
> _______________________________________________
> 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/20150114/67b06cff/attachment-0003.html>