[maker-devel] about predictor training

[熹熹]

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

[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/zff2augustus_gbk.pl

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

Thanks,

Carson

_______________________________________________ maker-devel mailing list maker...@box290.bluehost.com http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org