vcftools nucleotide diversity statistic (pi)

Question

[modi2020]

Dear fellows:

I know that Nei's Pi (nucleotide diversity statistic) is calculated per site using sequences belonging to more than one individuals. However, in vcftools, it seems like they can be calculated on a per individual bases (https://sourceforge.net/p/vcftools/mailman/message/29923488/). In another post by vcftools, it was defined as "the average number of differences between a pair of chromosomes" (https://sourceforge.net/p/vcftools/bugs/42/) which means to me that only one individual is enough for its calculation. We plan to use this method in a paper we are about to write since we don't have more than one individual per breed. Does anyone know the formula used in vcftools or if it is at all valid to calculate Pi in this way?

I would really appreciate your kind advice on this.

Answer 1

[caw5cv]

Nucleotide diversity is a population-level metric, the average number of differences between a pair of chromosomes, across all chromosome combinations within the population. If you only have one individual, maybe you're looking to report heterozygosity instead?

[modi2020]

Thank you so much caw5cv. I realize that its a population level metric. However, it seems like in vcftools its calculated on a per individual basis for SNPs in non-overlapping windows. Is that still useful if one had SNPs called on a single genome.