VEP failed to parse TBX_GENERIC error

Question

[bruce.moran]

Hi all,

I encountered an error running VEP (v81) which I have not encountered before. Error states:

E::get_intv] failed to parse TBX_GENERIC, was wrong -p [type] used?
The offending line was: "#chr   pos(1-based)    ref     alt ...

This is 0% into a 'calculating consequences' step. I am unsure if VEP has not annotated anything for chromosome it was operating on (although below the error is a 100% progress bar and time has been taken after this error). Also not sure what the -p [type] refers to(?)

My command is:

perl variant_effect_predictor.pl --force_overwrite --offline --cache \
--no_stats --everything --check_existing --total_length --allele_number \
--no_escape --gencode_basic --vcf --merged -custom /bin/GERP/All_hg19_RS.bw,GERP,bigwig \
--plugin dbNSFP,/bin/dbNSFP/dbNSFP.gz,1000Gp3_EUR_AF,ESP6500_EA_AF,SIFT_score,SIFT_pred,Polyphen2_HDIV_score,Polyphen2_HDIV_pred,LRT_score,LRT_pred,MutationTaster_score,MutationTaster_pred \
--pick_order biotype,tsl,canonical,length,ccds,rank \
--input_file $inp --output_file $out

which has worked for other VCF before.

Any thoughts are greatly appreciated, mainly I want to know if this is a tolerable error for single position, or more likely to be something requiring a fix and rerun.

Thanks,
Bruce.

[Pierre Lindenbaum]

what are is the VCF header of "$inp" ?

[bruce.moran]

VCF header of $inp is validated with GATK ValidateVariants which I trust to give me an error about even the most minor of issues!

[Emily]

Have informed our variation team about this and someone will be along to help you shortly.

Answer 1

[EnsemblWill]

The offending line reported is the header line from the dbNSFP data file.

If it continues to run (and you get the expected data back from dbNSFP) then you can probably ignore it.

If it's an error that persists and/or causes VEP to crash, try re-downloading and re-indexing the dbNSFP file according to the instructions in the plugin (https://github.com/ensembl-variation/VEP_plugins/blob/master/dbNSFP.pm)

Tabix should ignore any lines beginning "#" by default, so perhaps something has gone wrong here. It may be you need to create a dbNSFP file with a single header line at the top, and remove the header line from subsequent files that you concatenate on with cat:

head -n1 dbNSFP3.0a_variant.chr1 > dbNSFP.txt
cat dbNSFP3.0a_variant.chr* | grep -v "^#" >> dbNSFP.txt
bgzip dbNSFP.txt
tabix -s 1 -b 2 -e 2 dbNSFP.txt.gz

Saying that though, I haven't seen this error before, and tabix should ignore header lines.

[bruce.moran]

Thanks Will, thought it might be header related but then was unsure why it would occur around Chr6. So far only one instance but will take steps you recommend if it occurs again.

Answer 2

[Yunling]

Hi Will,

I did exactly as what you said and got the same error message:

perl /group/tools/vep/ensembl-tools-release-86/scripts/variant_effect_predictor/variant_effect_predictor.pl -cache --offline -i /group/tools/vep/ensembl-tools-release-86/scripts/variant_effect_predictor/example_GRCh37.vcf --dir_cache /scratch/scratch3/op_genomic2/transfer_drop/for-Mandy/.vep/ -force --plugin dbNSFP,dbNSFP.gz

2016-12-13 19:51:00 - Read existing cache info 2016-12-13 19:51:00 - Failed to instantiate plugin dbNSFP: ERROR: Could not read headers from dbNSFP.gz

2016-12-13 19:51:00 - Starting... 2016-12-13 19:51:00 - Detected format of input file as vcf 2016-12-13 19:51:00 - Read 173 variants into buffer 2016-12-13 19:51:00 - Reading transcript data from cache and/or database

It finished, but no dbNSFP results which are what I need.