Filtering VCF file by INFO flag

Question

[cl10101]

I am trying to filter variants with dbSNP annotation. In my VCF file this information is contained in INFO column like this:

Y       59003592        .       A       G       .       .       NS=1;AN=1;AC=1;CGA_XR=dbsnp.96|rs2140187;CGA_SDO=2      GT:PS:FT:GQ:HQ:EHQ:CGA_CEHQ:GL:CGA_CEGL:DP:AD:CGA_RDP        1:.:PASS:2035:2035,.:2035,.:44,.:-2035,0:-44,0:157:146,.:11

Information from header:

##INFO=<ID=CGA_XR,Number=A,Type=String,Description="Per-ALT external database reference (dbSNP, COSMIC, etc)">

According to vcftools documentation there is option to filter sites with specific INFO flag (--keep-INFO). I've tried to use this:

vcftools --vcf file.vcf --out output --keep-INFO CGA_XR

but without success:

VCFtools - 0.1.15
(C) Adam Auton and Anthony Marcketta 2009

Parameters as interpreted:
    --vcf file.vcf
    --out output
    --keep-INFO CGA_XR

After filtering, kept 1 out of 1 Individuals
Error: Using INFO flag filtering on non flag type CGA_XR will not work correctly.

What is the proper usage of this function?

Answer 1

[Shane McCarthy]

bcftools view -i 'CGA_XR ~"dbsnp"' file.vcf

Answer 2

[Pierre Lindenbaum]

vcffilterjs

 java -jar dist/vcffilterjs.jar -e 'variant.getAttributeAsString("CGA_XR","").startsWith("dbsnp")' file.vcf

Answer 3

[WouterDeCoster]

You could try:

cat <(grep '^#' yourfile.vcf) <(grep '|rs' yourfile.vcf) > outputfile.vcf

The first grep gets the header, the second grep every line containing a |rs pattern, which is hopefully specific enough. The result of both greps is cat together to create the output file.

[Pierre Lindenbaum]

grep -E "(^#|\|rs)" in.vcf > out.vcf

[WouterDeCoster]

That's a delicious solution!