custom "ab initio" predictions with automatic hint-based predictions

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

custom "ab initio" predictions with automatic hint-based predictions

Bob Zimmermann
Hello,

I would like to run maker with a custom set of ab initio predictions (based on hints given to augustus from RNAseq data), but allowing it to incorporate EST and protein data to make an additional run of augustus using hints derived from those alignments.

My gene prediction section of the maker_opts.ctl file looks like this:
...
augustus_species=all_combined #Augustus gene prediction species model
...
pred_gff=../ab_initio_predictions/all_combined.augustus_masked.gff3 #ab-initio predictions from an external GFF3 file
model_gff= #annotated gene models from an external GFF3 file (annotation pass-through)
est2genome=1 #infer gene predictions directly from ESTs, 1 = yes, 0 = no
protein2genome=0 #infer predictions from protein homology, 1 = yes, 0 = no


It seems as though even if pred_gff is set, augustus will still be run for ab initio predictions with no hints if an augustus_species setting is present. I was curious if there was any way around this, partly because custom ab initios could improve my annotation and also because the ab initio step can take long.

Thanks for your help!

Bob
_______________________________________________
maker-devel mailing list
[hidden email]
http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org
Reply | Threaded
Open this post in threaded view
|

Re: custom "ab initio" predictions with automatic hint-based predictions

Carson Holt-2
Hi Bob,

pred_gff is a way to get models MAKER cannot run into the analysis. Input to pred_gff will not get hints since MAKER is not running the program. Setting augustus_species allows MAKER to run Augustus with and without hints and then those models compete against each other. You cannot just run with hints as the raw model is also used as a filter to help reduce false positive gene models that result from bad hints. If the gff3 you are providing is the same as the MAKER run of Augustus, I would recommend not providing it. If it is different in some way, then you can leave it in.  If you run under MPI (it’s ok to run MPI on a single machine), then MAKER will parallelize the Augustus run by running multiple configs and contig chunks at the same time.

Thanks,
Carson





> On Oct 11, 2017, at 1:42 PM, Bob Zimmermann <[hidden email]> wrote:
>
> Hello,
>
> I would like to run maker with a custom set of ab initio predictions (based on hints given to augustus from RNAseq data), but allowing it to incorporate EST and protein data to make an additional run of augustus using hints derived from those alignments.
>
> My gene prediction section of the maker_opts.ctl file looks like this:
> ...
> augustus_species=all_combined #Augustus gene prediction species model
> ...
> pred_gff=../ab_initio_predictions/all_combined.augustus_masked.gff3 #ab-initio predictions from an external GFF3 file
> model_gff= #annotated gene models from an external GFF3 file (annotation pass-through)
> est2genome=1 #infer gene predictions directly from ESTs, 1 = yes, 0 = no
> protein2genome=0 #infer predictions from protein homology, 1 = yes, 0 = no
> …
>
> It seems as though even if pred_gff is set, augustus will still be run for ab initio predictions with no hints if an augustus_species setting is present. I was curious if there was any way around this, partly because custom ab initios could improve my annotation and also because the ab initio step can take long.
>
> Thanks for your help!
>
> Bob
> _______________________________________________
> maker-devel mailing list
> [hidden email]
> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org


_______________________________________________
maker-devel mailing list
[hidden email]
http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org
Reply | Threaded
Open this post in threaded view
|

Re: custom "ab initio" predictions with automatic hint-based predictions

Bob Zimmermann
Dear Carson,

Thanks for the response! Sorry for the slow reply.

Actually what I meant was that I wanted to generate other types of hints that maker could not automatically use to prevent lower quality ab initio predictions from influencing the final output. Therefore I wanted to make my own ab intio predicitions prior to running maker, and then have maker to generate the transcript hints and then run augustus, finally synthesizing my own ab initio predicions with the maker hint-based ones. (In other words, just run the second round of augustus, not the first one.)

I’ve attached a patch which seemed to allow me to tell maker to do what I wanted it to do. Am I missing something?

Best,
Bob



Department of Molecular Evolution and Development
Universität Wien
Althanstraße 14 (UZA I), Zimmer 2.019
1090 Vienna
Austria

+43 1 427757002




> On 13 Oct 2017, at 17:42, Carson Holt <[hidden email]> wrote:
>
> Hi Bob,
>
> pred_gff is a way to get models MAKER cannot run into the analysis. Input to pred_gff will not get hints since MAKER is not running the program. Setting augustus_species allows MAKER to run Augustus with and without hints and then those models compete against each other. You cannot just run with hints as the raw model is also used as a filter to help reduce false positive gene models that result from bad hints. If the gff3 you are providing is the same as the MAKER run of Augustus, I would recommend not providing it. If it is different in some way, then you can leave it in.  If you run under MPI (it’s ok to run MPI on a single machine), then MAKER will parallelize the Augustus run by running multiple configs and contig chunks at the same time.
>
> Thanks,
> Carson
>
>
>
>
>
>> On Oct 11, 2017, at 1:42 PM, Bob Zimmermann <[hidden email]> wrote:
>>
>> Hello,
>>
>> I would like to run maker with a custom set of ab initio predictions (based on hints given to augustus from RNAseq data), but allowing it to incorporate EST and protein data to make an additional run of augustus using hints derived from those alignments.
>>
>> My gene prediction section of the maker_opts.ctl file looks like this:
>> ...
>> augustus_species=all_combined #Augustus gene prediction species model
>> ...
>> pred_gff=../ab_initio_predictions/all_combined.augustus_masked.gff3 #ab-initio predictions from an external GFF3 file
>> model_gff= #annotated gene models from an external GFF3 file (annotation pass-through)
>> est2genome=1 #infer gene predictions directly from ESTs, 1 = yes, 0 = no
>> protein2genome=0 #infer predictions from protein homology, 1 = yes, 0 = no
>> …
>>
>> It seems as though even if pred_gff is set, augustus will still be run for ab initio predictions with no hints if an augustus_species setting is present. I was curious if there was any way around this, partly because custom ab initios could improve my annotation and also because the ab initio step can take long.
>>
>> Thanks for your help!
>>
>> Bob
>> _______________________________________________
>> maker-devel mailing list
>> [hidden email]
>> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org
>

_______________________________________________
maker-devel mailing list
[hidden email]
http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org

maker_noabinit.patch (970 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: custom "ab initio" predictions with automatic hint-based predictions

Carson Holt-2
Your patch will essentially just turn off all maker hint based gene prediction when no_abinit is turned on. We do not currently have a way to pass in external hints, but if you just want your hint based predictions to compete against MAKER hint based prediction, you can provide it as pred_gff while still letting MAKER run by giving the augustus_species file.

—Carson

> On Nov 28, 2017, at 7:37 AM, Bob Zimmermann <[hidden email]> wrote:
>
> Dear Carson,
>
> Thanks for the response! Sorry for the slow reply.
>
> Actually what I meant was that I wanted to generate other types of hints that maker could not automatically use to prevent lower quality ab initio predictions from influencing the final output. Therefore I wanted to make my own ab intio predicitions prior to running maker, and then have maker to generate the transcript hints and then run augustus, finally synthesizing my own ab initio predicions with the maker hint-based ones. (In other words, just run the second round of augustus, not the first one.)
>
> I’ve attached a patch which seemed to allow me to tell maker to do what I wanted it to do. Am I missing something?
>
> Best,
> Bob
>
> —
>
> Department of Molecular Evolution and Development
> Universität Wien
> Althanstraße 14 (UZA I), Zimmer 2.019
> 1090 Vienna
> Austria
>
> +43 1 427757002
>
> <maker_noabinit.patch>
>
>> On 13 Oct 2017, at 17:42, Carson Holt <[hidden email]> wrote:
>>
>> Hi Bob,
>>
>> pred_gff is a way to get models MAKER cannot run into the analysis. Input to pred_gff will not get hints since MAKER is not running the program. Setting augustus_species allows MAKER to run Augustus with and without hints and then those models compete against each other. You cannot just run with hints as the raw model is also used as a filter to help reduce false positive gene models that result from bad hints. If the gff3 you are providing is the same as the MAKER run of Augustus, I would recommend not providing it. If it is different in some way, then you can leave it in.  If you run under MPI (it’s ok to run MPI on a single machine), then MAKER will parallelize the Augustus run by running multiple configs and contig chunks at the same time.
>>
>> Thanks,
>> Carson
>>
>>
>>
>>
>>
>>> On Oct 11, 2017, at 1:42 PM, Bob Zimmermann <[hidden email]> wrote:
>>>
>>> Hello,
>>>
>>> I would like to run maker with a custom set of ab initio predictions (based on hints given to augustus from RNAseq data), but allowing it to incorporate EST and protein data to make an additional run of augustus using hints derived from those alignments.
>>>
>>> My gene prediction section of the maker_opts.ctl file looks like this:
>>> ...
>>> augustus_species=all_combined #Augustus gene prediction species model
>>> ...
>>> pred_gff=../ab_initio_predictions/all_combined.augustus_masked.gff3 #ab-initio predictions from an external GFF3 file
>>> model_gff= #annotated gene models from an external GFF3 file (annotation pass-through)
>>> est2genome=1 #infer gene predictions directly from ESTs, 1 = yes, 0 = no
>>> protein2genome=0 #infer predictions from protein homology, 1 = yes, 0 = no
>>> …
>>>
>>> It seems as though even if pred_gff is set, augustus will still be run for ab initio predictions with no hints if an augustus_species setting is present. I was curious if there was any way around this, partly because custom ab initios could improve my annotation and also because the ab initio step can take long.
>>>
>>> Thanks for your help!
>>>
>>> Bob
>>> _______________________________________________
>>> maker-devel mailing list
>>> [hidden email]
>>> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org
>>
>


_______________________________________________
maker-devel mailing list
[hidden email]
http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org