Doubt Regarding Combine Variants and Multisample SNP calling

anupamanupam Posts: 8Member

Hi sir,
I used GATK combine variants to call SNP from my 7 vcf files. Now I am using unified genotyper to call all samples simultaneously for SNP.
When I focused on my gene of interest (DLX6), in combine variants output, there are 5 "./." among 7 samples. but in case of multisample SNP calling, DLX6 gave 0/0 in those samples and its corresponding information. So, I want to know what is the meaning of "./." in combine variants results and which results should I take for analysis?

Best Answer


  • anupamanupam Posts: 8Member

    Hi sir,
    I have attached my gene of interest DLX6 results came from combine variants and Multisample SNP calling. please look the DLX6 variants information in 7 samples. It's a two line results, so won't take your much time.

    Thanks in advance.

  • anupamanupam Posts: 8Member

    Hi sir,
    Thank you for your reply. It clears my doubt regarding homozygous calls that are not coming in combine variants. But, there is some inconsistency in two of my samples. Younger son and wife sample contain heterozygous calls and satisfy other criteria (from multisample SNP calling),but it won't coming in combine variants results. I have attached my file and highlight the inconsistent samples in yellow. so please go through it once and let me know what's the reason for that inconsistency.

    Thank you for your time.

  • Geraldine_VdAuweraGeraldine_VdAuwera Posts: 9,963Administrator, Dev admin

    Ah, this is a good thing actually. This can happen if the site is difficult to genotype in a sample, and it illustrates the power of multisample calling to fix such problem. In single sample calling, there is not enough evidence to call the variant, so the program decides the sample is hom-ref at this site, and does not emit a variant call (so CombineVariants will assign ./.). However, in multisample calling, the program sees more than one sample with some evidence for the variant. It would not be enough evidence to call the variant individually per sample, but taken together it is enough to call the variant in the cohort, so the program emits a variant call.

    This shows you how using multisample calling can help you discover variants that you would not be able to see in single sample calling.

    Geraldine Van der Auwera, PhD

Sign In or Register to comment.