[maker-devel] Names/IDs from pred_gff are included in final gff

[Ole Kristian Tørresen]

Ok, but I have an entry in the final gff like this:
ID=maker-GmG20150304_scaffold_2371-pred_gff_maker-gene-0.125-mRNA-1;Parent=maker-GmG20150304_scaffold_2371-pred_gff_maker-gene-0.125;Name=ENSGMOT00000000668.1;_AED=0.00;_eAED=0.00;_QI=819|1|1|1|1|1|4|112|726;score=89.75616

(The name is derived from a pred_gff entry which is the results of mapping an old annotation to the new assembly). 

This is then called ENSGMOT00000000668.1 protein AED:0.00 eAED:0.00 QI:819|1|1|1|1|1|4|112|726 in the proteins.fasta file. Which is unfortunate, because it apparently mapped 12 places in the assembly.

I have set map_forward=0, but keep_preds=1 (filtering on domain presence and AED score later). 

This and another file (result of genemark_gtf2gff3), is not input as match/match_part to MAKER, but with gene/exon/CDS/mRNA. Could that be the issue?

Ole

 
> On 16 Jan 2018, at 19:15, Carson Holt <carsonhh at gmail.com> wrote:
> 
> pred_gff will maintain it’s name in the match/match_part feature as the information is pulled directly from the input GFF3. But any pred_gff feature that becomes a final model will be renamed to something like “scaffold_1517-pred_gff_GeneMark.hmm-gene-0.6-mRNA-1” unless you specify map_forward=1 to maintain the original name.
> 
> —Carson
> 
> 
> 
>> On Jan 4, 2018, at 6:21 AM, Ole Kristian Tørresen <o.k.torresen at ibv.uio.no> wrote:
>> 
>> Hi,
>> as far as I can see, names or IDs of features in gffs given to pred_gff is included in the final output as the name of the feature. As far as I can understand, this is not expected behaviour (it is for model_gff). This is with MAKER 2.31.9.
>> 
>> I have these settings:
>> map_forward=0 
>> keep_preds=1 
>> 
>> I thought that map_forward had to be 1 to get the names for the old GFFs. Can you replicate this?
>> 
>> Thank you.
>> 
>> Sincerely,
>> Ole K. Tørresen
>> 
>> _______________________________________________
>> maker-devel mailing list
>> maker-devel at box290.bluehost.com
>> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org
>