exonerate Protein alignments in Braker1

Discussions about BRAKER, the automated joint usage of GeneMark and AUGUSTUS.

Moderator: bioinf

Post Reply
cjacobs
Posts: 1
Joined: Tue Jun 07, 2016 12:40 pm

exonerate Protein alignments in Braker1

Post by cjacobs »

Dear Katharina,

I am trying to use braker by using both protein alignment evidence and RNAseq data. As I understand it it should be possible. I do have some questions about it though.

I've run exonerate with the protein2genome model to align multiple proteomes to the genome. Next I simply pasted these different alignments into one file using cat:
cat protein2genome1.gff protein2genome2.gff etc. > protein2genome_combined.gff

Next I used the script included with augustus 3.2.1 to convert them into hints.

exonerate2hints.pl --in=protein2genome_combined.gff --source=P --out=exonerate.gff

So my question would be, can I simply add these different aligned proteomes together or won't this work?
Also, the extrinsic.cfg file included in the augustus config directory states that the file should include source=P
The file contains src=P , will that work too or should I change it?

My next question is, can I combine RNAseq and protein alignment data in Braker? I have mapped RNAseq data from different experiments with Hisat into the resulting accepted_hits.bam file. To combine the RNAseq and protein alignment data I plan to run braker like this:

braker.pl --softmasking=1 --extrinsicCfgFile=extrinsic.MP.cfg --hints=exonerate.gff --genome=genome.fa --species=speciesname --bam=accepted_hits.bam

So should that work or should I run braker with only the RNAseq data first and then run augustus seperately with the protein data? Also, will the standard version of extrinsic.MP.cfg work as I am not sure what would be good values otherwise.

One more question, do you believe that it is worth it to run several gene predictions and combine the predictions using evigan? Especially because braker already includes several lines of evidence. Not sure whether running for example augustus separately without hints (but using the config created in braker), running ballgown prediction and perhaps use some more predictors and combining it with evigan would actually improve the predictions.

Thank you for your input!

Best regards,

Chris Jacobs
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: exonerate Protein alignments in Braker1

Post by katharina »

It is currently not possible to use protein alignments with BRAKER1, yet. We are working on a new release that will enable you to do exactly this, but it is not out, yet.
Post Reply