Exac and gnomad MAF value in VCF

Question

[Sharon]

I want to know the difference between zero and '.' for an Exac value/GnomAD MAF value in an annotated VCF ? What is the difference between MAF is 0 as not found in the population and MAF is '.' as doesn't exist? Like for example, when the variant ExAC All MAF value is '.' what does this mean? I have lots of variants with this MAF in ExAC and GnomAD value both in VEP and Annovar annotated VCFs.

Thanks

Answer 1

[Pierre Lindenbaum]

0 is zero.

. is for null: data is not available/ unknown /missing

[Sharon]

Yes, my professor is asking me a confusing question for me. What does null mean? Does it mean not found in the population, so why not it is counted as zero then. Does null mean not sequenced, hard to say this. Does it mean not known by the annotator like VEP or Annovar? Like what missing means, missing from population or what?

[WouterDeCoster]

For a more general explanation on the difference between zero and null: https://softwareengineering.stackexchange.com/a/134872

But as Pierre already said, the dot means that the data is missing. It's unknown whether it is actually zero. It's Schrodinger's MAF: you don't know what it is if it's not available or calculated. Perhaps an unreliable region, in which variant calls were not performed.

Adding an example to your question is always helpful...

[Sharon]

I understand the difference between zero and null in computer science/software engineering, that's not my question. I mean zero and null in the variant annotations context. A zero Maf means not found in the population. Null means missing. What is the difference between not found and missing. Example: Exac value below is '.' not zero

Chr Start   End Ref Alt Func.refGene    Gene    ExAC_ALL

1   117150628   117150628   G   C   exonic  IGSF3        .

[Pierre Lindenbaum]

let's say that the coverage was very bad in GNOMAD for this region: there was NO read overlapping this region, while the variant was found with another experiment (that's why you have this variant).So you have MAF_GNOMAD=AC/AN with AN=0 (because there is no allele at all). So you would put an MAF_GNOMAD=.. On the other side, the sequencing was ok in gnomad (AN>0) but the alt allele was not found in the population (AC==0), then you would have MAF_GNOMAD=0

[Sharon]

Got it, thanks all so much !