Hey there,

I am trying to randomly select variants from a multisample vcf file which I generated by merging several individual vcf files together using bcftools merge.

This is the command that I am using:

gatk SelectVariants -R /path/to/ref.fasta -V input.vcf -O output_1pct.vcf --select-random-fraction 0.01--exclude-non-variants true --exclude-filtered true

Unfortunately, it does not select 1 percent of the variants but instead outputs a file with only one variant which is always the same. I tried running this command without the optional arguments but it still outputs this variant:

#CHROM  POS ID  REF ALT QUAL    FILTER  INFO    FORMAT  indivA  indivB  indivC  indivD  indivE  indivF  indivG  indivH  indivI  indivJ  indivK  indivL  indivM  indivN
35  964 .   A   T   141.05  PASS    AC=1;AN=2   GT:AO:DP:PL:QA:QR:RO    ./. ./. ./. ././.   0/1:6:11:184,0,151:237:200:5    ./. ./. ./. ./. ./. ./. ./. ./.

Also, when I set --select-random-fraction to 1 (100%) it only selects 34 variants although in reality there are 34870672!! It just picks the 34 first variants...

ValidateVariants doesn't give any error messages on the multisample vcf.

However, when I run this same command on an individual vcf it seems to work. Any idea what is going wrong?

GATK version : GATK/4.1.1.0
java version: java/sun_jdk1.8.0_151

Thanks in advance!