"expected GQ field in snps VCF" while (I think !) my vcf contains FORMAT/GQ

[lindenb]

Hi Brent,

I'm running duphold for the 1st time with snps.vcf computed with bcftools mpileup+call.
My SV vcf was computed with manta.

I'm getting the following error:

+ /ccc/work/cont007/fg0019/lindenbp/packages/duphold/duphold --snp /ccc/scratch/cont007/fg0113/lindenbp/20190428.TROUCARD.MANTA/work/bd/c70293fb517483436faf17dd3b2e61/B00I4BY.snv.bcf --vcf jeter.bcf --bam /ccc/genostore/cont007/fg0073/fg0073/analyse/projet_TROUCARD_605/ANALYSE/ANALYSE_T605_B00I4BY_HJWJJALXX_2_hs37d5/MAPPING_B00I4BY/reliable.realign/T605_DA_B00I4BY_HJWJJALXX_hs37d5_MERGE_PE_2.reliable.realign.bam --fasta /ccc/work/cont007/fg/fg/biobank/by-taxonid/9606/hs37d5/hs37d5_all_chr.fasta --threads 4 -o jeter.duphold.vcf
I, [2019-05-01T14:48:04] -- duphold: starting to read snps for chrom: chr1
expected GQ field in snps VCF

1. unless I'm wrong, my snv.bcf do contains the FORMAT/GQ field:

$ bcftools view /ccc/scratch/cont007/fg0113/lindenbp/20190428.TROUCARD.MANTA/work/bd/c70293fb517483436faf17dd3b2e61/B00I4BY.snv.bcf | grep -v '##' | cut -f 9 | sort | uniq
FORMAT
GT:PL:DP:AD:GQ
$

what am I doing wrong ?

2. it seems that for this error, duphold doesn't exit with failure/-1 because the next statement was executed while I'm running with set -e

3. minor: in the doc, it should be specified that snp.bcf should be indexed.

thank your for your help

[brentp]

Hi Pierre, this was a logic error I introduced last release. Please try this binary (gunzip and chmod+x)
duphold.gz

and I'll fix this for the next release.

[lindenb]

thanks !

[brentp]

also, this should have resulted in a non-zero exit code so I'm not sure what happened there.

[lindenb]

yes, i tested it later, there is something weird on my side

[lindenb]

ah, I deleted the wrong part of my minor comment .

Mnior: in README.md there is a small problem of open 'single quote' vs close 'double quote' in

bcftools view -i '(SVTYPE = "DEL" & FMT/DHFFC[0] < 0.7) | (SVTYPE = "DUP" & FMT/DHBFC[0] > 1.3)" $svvcf

[brentp]

ok. I pushed a fix for this, including in the readme. Thanks much for reporting.