SNP data - significance heterozygosity test?

Question

[Roland]

I have a dataset of 20 individuals that has been genotyped at 25000 SNPs. From this data, I've calculated the mean observed and expected heterozygosity values, as well as standard errors. But how do I test if there's a signifiance difference between the observed and heterozygosity means? Do I just do a simple t-test?

Thanks.

Answer 1

[Fabio Marroni]

Several software tools can assess deviations from Hardy-Weinberg proportions for you. This test is actually testing the difference between expected (expected under HW equilibrium) and observed heterozygosity.

For each SNP with alleles A and a you can easily perform the calculation by hand by comparing the observed number of AA, Aa and aa genotypes with the expected number of the same genotypes and then test using a chi-square distribution with 2 degrees of freedom.