[maker-devel] about predictor training

[Carson Holt]

You need to convert the GTF files to GFF3.  There is a tophat2gff and
cufflinks2gff script that come with MAKER.  I recommend only using cufflinks
results and ignoring tophat results though as they tend to be a lot more
spurious.

Jason Stajich wrote an excellent explanation on training Augustus on the
list previously - 
http://brie4.cshl.edu/pipermail/gmod-help/2012-June/001724.html
He also included scripts to assist with the training -
https://github.com/hyphaltip/genome-scripts/blob/master/gene_prediction/zff2
augustus_gbk.pl

Overall the strategy is similar to the one used to train SNAP.

Thanks,
Carson


From:  熹熹 <linzkl007 at hotmail.com>
Date:  Saturday, 11 May, 2013 1:28 PM
To:  "maker-devel at yandell-lab.org" <maker-devel at yandell-lab.org>
Subject:  [maker-devel] about predictor training

Hi,

I'm trying to use MAKER to annotate the new genome sequence which I
assembled by myseft. I used TopHat and Cufflinks to align the sequence based
on the RNA-seq we have. Based on the tutorial of MAKER, I may need three
fasta format file including assembly data, ESTs and protein database to
train the SNAP. I may use SwissProt as the protein database. Can I use the
gtf result from Cufflinks directly as an ESTs during the training?
Another is, if I want to use Augustus to do the ab initio gene prediction,
do I need to do the same way as SNAP? Cause I saw some posts that the result
from ab initio would be used as the evidence to train the predictor. Can I
ask is there has some order doing the prediction in different predictor?

Thank you so much for you help.

Lin
       
_______________________________________________ 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/20130513/dec7d24f/attachment-0003.html>