Computational protocol: HLA class I haplotype diversity is consistent with selection for frequent existing haplotypes

Similar protocols

Protocol publication

[…] In the Ewens-Watterson test, one calculates the hypothetical F1 homozygosity following random mating (the sum of squared allele frequencies) of the sample, and compares it to the expected homozygosity of a sample with the same attributes (sample size and total number of alleles) from the Ewens sampling formula. The expected homozygosity and the p-value of the test are usually obtained using Slatkin's method []. The test parameters are limited by the numerical calculation of Ewens' sampling formula so that the maximal number of alleles in a sample is currently limited to a thousand and the maximal sample size to a few thousand. In order to perform such tests for extremely large samples such as ours, a representative subsample must be taken.We randomly sampled 1,200 haplotypes from each US subpopulation and performed the EW test on the subsample, using either the Arlequin [] or PyPop [] software packages with similar results. We calculated Fnd values for each sample (Fnd = (Fobs − Fexp)/Var(Fexp)1/2), where Fobs is the observed homozygosity of the sample, and Fexp is the expected homozygosity for a population with the same parameters calculated using random samples, which also provided an estimate for the variance. […]

Pipeline specifications

Software tools Arlequin, PyPop
Application Population genetic analysis
Organisms Homo sapiens