EVM with no ab initio models

[Paolo]

Dear EVM users,

I am dealing with a genome assembly in which I struggle to get meaningful ab initio gene prediction (with e.g. Augustus). I was wondering whether I can combine all other evidences I have collected so far (PASA transcript alignments, exonerate and genewise protein alignments, stringtie/transdecoder gene prediction) in EVM without any ab initio gene model.

I gave a try setting the stringtie/transdecoder evidence as "OTHER_PREDICTION" in the weight file, EVM completed the run and I got a genome annotation with a reasonable number of genes, but I am not sure if I can trust it. Can I please have your feedback on this?

Thanks for your help!

Paolo

[Brian Haas]

I don't think that's going to work w/ EVM.

PASA/StringTie/TransDecoder will probably get you most of what you're after, assuming your rna-seq data is comprehensive enough.

--
You received this message because you are subscribed to the Google Groups "EVidenceModeler-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to evidencemodeler-...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/evidencemodeler-users/28fc6a3e-edb2-4c7c-af9c-11f6e9adab31%40googlegroups.com.

--

--
Brian J. Haas
The Broad Institute
http://broadinstitute.org/~bhaas

 

[Paolo]

Dear Brian,

Thanks for the quick reply. My transcriptome is quite comprehensive as I have BUSCO scores exceeding 90%, but I was thinking about using EVM because I wanted to include the protein alignment information. Any suggestion on how I can include it?

Cheers

On Tuesday, November 12, 2019 at 5:27:26 PM UTC+1, Brian Haas wrote:

I don't think that's going to work w/ EVM.

PASA/StringTie/TransDecoder will probably get you most of what you're after, assuming your rna-seq data is comprehensive enough.

On Tue, Nov 12, 2019 at 11:23 AM Paolo <paolo.fra...@gmail.com> wrote:

Dear EVM users,

I am dealing with a genome assembly in which I struggle to get meaningful ab initio gene prediction (with e.g. Augustus). I was wondering whether I can combine all other evidences I have collected so far (PASA transcript alignments, exonerate and genewise protein alignments, stringtie/transdecoder gene prediction) in EVM without any ab initio gene model.

I gave a try setting the stringtie/transdecoder evidence as "OTHER_PREDICTION" in the weight file, EVM completed the run and I got a genome annotation with a reasonable number of genes, but I am not sure if I can trust it. Can I please have your feedback on this?

Thanks for your help!

Paolo

--
You received this message because you are subscribed to the Google Groups "EVidenceModeler-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to evidencemodeler-users+unsub...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/evidencemodeler-users/28fc6a3e-edb2-4c7c-af9c-11f6e9adab31%40googlegroups.com.

[Brian Haas]

Sure thing.  I'd suggest working with the transcript data first, then have a look at cases where you think the protein alignment data could have been helpful towards arriving at an improved structure annotation. If you find enough convincing cases, then there might be a path forward toward trying to push things further.  It might involve leveraging PASA to do it.  

The thing you want to avoid is capturing more protein alignment-based structures, but ruining others where good transcript evidence exists.  There are also issues with protein alignment evidence highlighting pseudogenes, and you'd want to avoid integrating these as potential true coding genes.

On Tue, Nov 12, 2019 at 11:38 AM Paolo <paolo.fra...@gmail.com> wrote:

Dear Brian,

Thanks for the quick reply. My transcriptome is quite comprehensive as I have BUSCO scores exceeding 90%, but I was thinking about using EVM because I wanted to include the protein alignment information. Any suggestion on how I can include it?

Cheers

On Tuesday, November 12, 2019 at 5:27:26 PM UTC+1, Brian Haas wrote:

I don't think that's going to work w/ EVM.

PASA/StringTie/TransDecoder will probably get you most of what you're after, assuming your rna-seq data is comprehensive enough.

On Tue, Nov 12, 2019 at 11:23 AM Paolo <paolo.fra...@gmail.com> wrote:

Dear EVM users,

I am dealing with a genome assembly in which I struggle to get meaningful ab initio gene prediction (with e.g. Augustus). I was wondering whether I can combine all other evidences I have collected so far (PASA transcript alignments, exonerate and genewise protein alignments, stringtie/transdecoder gene prediction) in EVM without any ab initio gene model.

I gave a try setting the stringtie/transdecoder evidence as "OTHER_PREDICTION" in the weight file, EVM completed the run and I got a genome annotation with a reasonable number of genes, but I am not sure if I can trust it. Can I please have your feedback on this?

Thanks for your help!

Paolo

--
You received this message because you are subscribed to the Google Groups "EVidenceModeler-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to evidencemodeler-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/evidencemodeler-users/28fc6a3e-edb2-4c7c-af9c-11f6e9adab31%40googlegroups.com.

--

--
Brian J. Haas
The Broad Institute
http://broadinstitute.org/~bhaas

 

--

You received this message because you are subscribed to the Google Groups "EVidenceModeler-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to evidencemodeler-...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/evidencemodeler-users/11134b5a-f4d3-4346-b956-9c9f5dd88ccd%40googlegroups.com.