[Correct statistical analysis of genotype frequencies for UCP and PPAR gene families in residents of besieged Leningrad and the control group].

Correct harmonized statistical re-analysis of the data published in this Journal by I.V.Polyakova et al. (2014) clearly shows that, contrary to the authors' opinion, the distribution of genotypes among residents of besieged Leningrad and the residents of the North-West region of Russia appeared to be statistically indistinguishable in all five genes studied. The main causes of the erroneous conclusions of the authors are neglecting the problem of multiple comparisons and fundamental impossibility of sampling adequate control group. A scheme for harmonized statistical analysis of such data is presented. It implies not only frequentist but Bayesian point and interval estimates for genotype proportions and their differences, for fixation index (coefficient of inbreeding) FIS, for the effect size φ based on χ2 statistic (contingency coefficient) and for the achieved power (1 - β), as well as estimates of posterior probabilities for the null hypothesis P(H_0 |D), Bayes factors 〖BF〗_01, observed p-values, p_obs, with the prediction intervals, and p-values adjusted for the multiplicity of null hypotheses tested (P_S).