Why call variant generate homozygous site with genotype 0/0, why it be reported even if it same with reference genome?

Question

[Nich]

Hi, guys I use gatk to call variant and get a vcf file. When I look into it, I find some homozygous site with genotype 0/0.

chr2    231861039       .       CCT     .       793     .       AN=2;DP=275;MQ=60.00    GT:AD:DP        0/0:275:275

I think this variant means it same with reference genome. So I don't know why it still been reported by SNP caller software?

[Pierre Lindenbaum]

what was the command line / version ?

[Nich]

gatk gatk-4.1.0.0 HaplotypeCaller

/software/gatk4.0/gatk-4.1.0.0/gatk --java-options    '-Xmx20g'  HaplotypeCaller  -L  region.bed -R  ucsc.hg19.fasta -I ccr.bam  -O ccr.vcf     >ccr.std 2>&1

nothing special

[jean.elbers]

Is this multiple sample variant calling? If so, then perhaps other samples have 0/1 and 1/1 genotypes if this is a biallelic SNP.

[Nich]

I just use gatk HaplotypeCaller call one sample

[jean.elbers]

Hmmm...well, after looking at your example output, it even shows there is no ALT allele. You could get rid of rows with like that with awk '$5!="."' OFS='\t' input-vcf-file > new-vcf-file-without-false-variants or something like that.

Answer 1

[d-cameron]

Reporting homozygous reference is important as it distinguishes that position from a no call position.

To answer the question: do samples A and B have the same alleles at position X, you need to know what the alleles are in both samples. If your SNV caller is unable to call position X in one of the samples (e.g. insufficient enough coverage, read mapqs too low, ...), then the answer to that question is "I don't know".This is not the same as "hom ref in both" and "ref in one, non-ref in other" and these cases need to be treated differently.