This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Format in vcf files in Geuvadis data not recognized

Hi,

I have downloaded Geuvadis genotypes from:

https://www.ebi.ac.uk/arrayexpress/files/E-GEUV-1/GEUVADIS.chr22.PH1PH2_465.IMPFRQFILT_BIALLELIC_PH.annotv2.genotypes.vcf.gz

And I am trying to recreate vcf files with only the European samples using bcftools, but when I try

bcftools view --samples-file 373_sampleIDs.tab GEUVADIS.chr22.PH1PH2_465.IMPFRQFILT_BIALLELIC_PH.annotv2.genotypes.vcf.gz

I get this error:

 [W::vcf_parse_format] FORMAT 'PP' is not defined in the header, assuming Type=String
 [W::vcf_parse_format] FORMAT 'BD' is not defined in the header, assuming Type=String
Undefined tags in the header, cannot proceed in the sample subset mode.

Trying to fix it with bcftools reheader does not work:

bcftools view -h file.vcf > header.txt; bcftools reheader -h header.txt file.vcf > fixed.vcf

I can find information on the FORMAT "PP", but not on the FORMAT "BD". 

##FORMAT=<ID=PP,Number=G,Type=Integer,Description="Phred-scaled Posterior Genotype Probabilities">

Would anybody be able to help? Thanks!

samtools format vcf bcftools geuvadis
• 3,519 views • link
updated by Pierre Lindenbaum Pierre Lindenbaum • written by jamespower • jamespower

W::

This is a warning, not an error. If the results is what you expect then everything is okay.

• 1 views • link
written by WouterDeCoster WouterDeCoster

Even if it says it as a warning, I cannot do certain operations such as extracting sample IDs (I update the last output for the error message).

• 1 views • link
written by jamespower • jamespower

Hello jamespower,

bcftools is very strict about header informations. So if you want to use it, you must fixe this information or use other programs that are not that strict like snpSift.

The header you've found for PP says it is from the type Integer. But in the message you show before it expect a String.

Have a look into the vcf specifation and modify your header, so that the missing fields are in and have the right type. Also make sure that the value for Number is correct.

If you show us the complete header of your current file and the first variants, we could help you doing this.

fin swimmer

• 1 views • link
written by finswimmer finswimmer

1 answer

the following command seems to fix the error. But I've no idea about the description of the new FORMAT. There are also some extra spaces in the INFO column...

 wget -O - "https://www.ebi.ac.uk/arrayexpress/files/E-GEUV-1/GEUVADIS.chr22.PH1PH2_465.IMPFRQFILT_BIALLELIC_PH.annotv2.genotypes.vcf.gz" |\
gunzip -c |\
awk '/^#CHROM/ {printf("##FORMAT=<ID=BD,Number=1,Type=String,Description=\"?\">\n##FORMAT=<ID=PP,Number=1,Type=String,Description=\"?\">\n");} {print}' |\
sed 's/ damaging/_damaging/g'  > fixed.vcf
• 1 views • link
written by Pierre Lindenbaum Pierre Lindenbaum

Log in to answer this question.

More posts like this