Background: Relationship study between disease and gene in case-control studies on the basis of allele frequency requires the existence of Hardy-Weinberg equilibrium in a population. If we ignore deviation from HWE, dependent on the value of the deviation, the true type-one-error rate will be increased or decreased.

Methods: The study used an expectation-maximization algorithm for estimate allele frequency with deviation from Hardy-Weinberg equilibrium. To illustrate, we applied data of type II diabetes and polymorphism of SLC30A8 that were published by Mohaddes et al.  This data is diallelic C and T that C is dominated.

Results: Mohaddes et al. not detected significant association between rs13266634 variant of SLC30A8 gene and T2D in our study population  while we detected significant association between them using EM-algorithm  In addition, the result of this study shows type-one-error rate for association test based on EM algorithm estimation of allele frequency is stable compared to the amount of deviation, although the true rate is slightly larger than the nominal value.

Conclusion: Our results show deviation from HWE is important in association studies and ignore it due to miss results.

genetic association studies, EM-algorithm, SLC30A8 gene, diabetes.