Das (2006) has pointed out some ‘serious analytical and interpretational flaws’ in our article.2 We admit that we had incorrectly stated that all the three loci (Thr394Thr, Gly482Ser and A2962G) are in Hardy–Weinberg equilibrium (HWE) among both cases (Type 2 diabetes) and controls (glucose-tolerant individuals). We have mistakenly compared the observed value of the statistic with the tabulated value of the χ2 distribution with 2 degrees of freedom, instead of 1 degree of freedom. This resulted in higher P-values and our incorrect inference that genotype frequencies were not significantly different from those expected under HWE. Indeed, as correctly pointed out by Das, all the three loci show statistically significant departures from HWE among cases. Das has also stated that the A2962G polymorphism is not in HWE among controls. His calculated P-value (0.006) is incorrect; the correct value is 0.02. We note that we have performed six independent tests of HWE (two each for the three polymorphisms), a P-value of 0.02 after correcting for multiple comparisons does not provide statistically significant evidence of departure from HWE. Thus, departures from HWE are only observed among cases, which as discussed later are not unexpected and do not invalidate the inferences reported in our paper.
Das (2006) has suspected that genotyping errors are responsible for the observed statistically significant deviations from HWE. Although, in principle, this may be true, we believe that it is unlikely for our sample. The reasons are: (a) for each locus, we have carried out duplicate genotypings in a randomly selected subset of 20% individuals and did not find discrepancies, (b) for each locus, all variant genotypes (both homozygotes and heterozygotes for the variant allele) were confirmed by mini-resequencing using an automated DNA sequencer (ABI 310 Genetic Analyzer); and (c) if indeed erroneous genotyping was the cause of departure from HWE, then it is unlikely that the errors would be confined to the case samples and not the control samples. Therefore, the probable reason for departure of observed genotype frequencies among cases for the three loci is not genotyping error but natural selection and linkage-disequilibrium. For any locus that is in strong linkage disequilibrium (LD) with a causal locus for the disease under study, departure from HWE is expected among cases who are likely to have reduced fitness.
This is a preview of subscription content, access via your institution
