[ensembl-dev] VEP variants missing on output

Will McLaren wm2 at ebi.ac.uk
Tue May 21 16:13:22 BST 2013


You get one line of output for each variant/feature overlap, so you will
almost always see more output lines than input if you use the default
output format. If you use VCF output, you only get one line per variant.

You can check how many unique variants there are in the output with e.g.:

grep -v # variant_effect_output.txt | cut -f 1 | sort -u | wc -l

assuming your variants have unique names.

Try dropping "html" from your config, see if that makes any difference - as
the newest feature there, it's got a higher chance of causing problems!

Will




On 21 May 2013 16:02, Guillermo Marco Puche <
guillermo.marco at sistemasgenomicos.com> wrote:

>  Hello Will,
>
> I'm getting more 3000 lines of file output.. this seems really weird....
>
> wc -l variant_effect_output.txt
>
> *3936*
>
> Here's the way I'm proceeding:
>
> ./variant_effect_predictor.pl -i /home/likewise-open/SGNET/gmarco/vep_71_annotation_check/input.vcf -force -fork 4 --database --config vep_71.test
>
>
> Here's the content of vep_71.test:
>
> dir                /home/likewise-open/SGNET/gmarco/.vep
> toplevel_dir       /home/likewise-open/SGNET/gmarco/.vep
> force_overwrite    1
> format             vcf
> html               1
> host               192.19.x.xx
> port               3306
> user               myuser
> password           mypassword
> buffer_size        5000
>
> hgvs               1
> canonical          1
> ccds               1
> check_svs          1
> domains            1
> gmaf               1
> hgnc               1
> maf_1kg            1
> numbers            1
> polyphen           b
> regulatory         1
> sift               b
>
> Best regards,
> Guillermo.
>
>
> On 05/21/2013 02:30 PM, Will McLaren wrote:
>
> Hi Guillermo,
>
> I'm unable to recreate this, sorry!
>
>  I get 406 going in, 406 coming out every time, whichever combination of
> those options above I use, and whether I use VCF or standard output.
>
>  Here's my run (minus -check_sv):
>
>  > perl variant_effect_predictor.pl -i guill.vcf -vcf -cache -force -fork
> 4 -hgvs -canon -ccds -domains -gmaf -hgnc -maf_1kg -numbers -poly b -regu
> -sift b -fasta ~/NFS/Fasta/Homo_sapiens.GRCh37.69.dna.primary_assembly.fa
> 2013-05-21 13:24:26 - Checking/creating FASTA index
> 2013-05-21 13:24:26 - Read existing cache info
> 2013-05-21 13:24:26 - Starting...
> 2013-05-21 13:24:26 - Detected format of input file as vcf
> 2013-05-21 13:24:26 - Read 406 variants into buffer
> 2013-05-21 13:24:26 - Reading transcript data from cache and/or database
> [================================================================]  [ 100%
> ]
> 2013-05-21 13:24:30 - Retrieved 10891 transcripts (0 mem, 10919 cached, 0
> DB, 28 duplicates)
> 2013-05-21 13:24:30 - Reading regulatory data from cache and/or database
> [================================================================]  [ 100%
> ]
> 2013-05-21 13:24:35 - Retrieved 36955 regulatory features (0 mem, 36955
> cached, 0 DB, 0 duplicates)
> 2013-05-21 13:24:35 - Calculating consequences
> [================================================================]  [ 100%
> ]
> 2013-05-21 13:24:56 - Writing output2013-05-21 13:24:56 - Processed 406
> total variants (14 vars/sec, 14 vars/sec total)
> 2013-05-21 13:24:56 - Wrote stats summary to
> variant_effect_output.txt_summary.html
> 2013-05-21 13:24:56 - Finished!
> > wc -l variant_effect_output.txt
> 408
>
>  It's 408 as it's adding two header lines to the VCF output.
>
>  Which 16 are missing from your output, and is it the same 16 each time?
>
>  Try writing to a different output file, or on a different disk if you
> can (perhaps disk space is an issue?)
>
>  Will
>
>
> On 21 May 2013 13:15, Guillermo Marco Puche <
> guillermo.marco at sistemasgenomicos.com> wrote:
>
>>  Hello Will,
>>
>> Here's the input:
>> https://github.com/guillermomarco/vep_plugins_71/blob/master/missing_variants/missing_output_variants.vcf
>>
>> As you said it's not about the options or plugins. Launching VEP without
>> specyfiying any option still returns an output with missing variants.
>>
>> Regards,
>> Guillermo.
>>
>>
>>
>> On 05/21/2013 01:49 PM, Will McLaren wrote:
>>
>> Hi Guillermo,
>>
>>  None of those options should filter out variants.
>>
>>  Are you able to provide any of the files that recreate the problem?
>>
>>  Is there any chance that you are using VCF input and it contains
>> non-variant lines - this would be where the ALT column is empty or "."? If
>> so, this may be your problem. To force these to be included in the output,
>> you should add --allow_non_variant.
>>
>>  Regards
>>
>>  Will
>>
>>
>> On 21 May 2013 09:40, Guillermo Marco Puche <
>> guillermo.marco at sistemasgenomicos.com> wrote:
>>
>>>  Hello,
>>>
>>> I've been checking VEP results, and i've noticed that I'm missing some
>>> input variants in the output.
>>>
>>> I think this may be issued to some of the options i'm using to launch
>>> vep:
>>>
>>> hgvs               1
>>> canonical          1
>>> ccds               1
>>> check_svs          1
>>> domains            1
>>> gmaf               1
>>> hgnc               1
>>> maf_1kg            1
>>> numbers            1
>>> polyphen           b
>>> regulatory         1
>>> sift               b
>>>
>>> Should be any of these options filtering output? I've disabled all
>>> plugins to run this test to be sure that it's not a plugin issue.
>>>
>>>
>>>    - With a 406 variant input vcf file, I've missed 16 variants.
>>>     - I then ran VEP with only those 16 missing variants and missed 3
>>>    on output.
>>>     - Rerun again and now with 3 missing variants and now not a single
>>>    one was missing.
>>>
>>> I would like to know what's behind that weird behaviour.
>>>
>>> Thank you.
>>>
>>> Best regards,
>>> Guillermo.
>>>
>>>
>>>
>>> _______________________________________________
>>> Dev mailing list    Dev at ensembl.org
>>> Posting guidelines and subscribe/unsubscribe info:
>>> http://lists.ensembl.org/mailman/listinfo/dev
>>> Ensembl Blog: http://www.ensembl.info
>>
>>
>> _______________________________________________
>> Dev mailing list    Dev at ensembl.org
>> Posting guidelines and subscribe/unsubscribe info:
>> http://lists.ensembl.org/mailman/listinfo/dev
>> Ensembl Blog: http://www.ensembl.info/
>>
>>
>
>
> _______________________________________________
> Dev mailing list    Dev at ensembl.org
> Posting guidelines and subscribe/unsubscribe info: http://lists.ensembl.org/mailman/listinfo/dev
> Ensembl Blog: http://www.ensembl.info/
>
>
>
>
> _______________________________________________
> Dev mailing list    Dev at ensembl.org
> Posting guidelines and subscribe/unsubscribe info:
> http://lists.ensembl.org/mailman/listinfo/dev
> Ensembl Blog: http://www.ensembl.info/
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ensembl.org/pipermail/dev_ensembl.org/attachments/20130521/34217cc6/attachment.html>


More information about the Dev mailing list