Nam J M, Gart J J
Am J Hum Genet. 1987 Jul;41(1):70-6.
Stevens suggests a test of fit, based on Bernstein's estimators, of the Hardy-Weinberg law for the ABO system. Nam and Gart extend this test to the generalized ABO-like system and apply it to HLA data. When the recessive gene is rare, Huether and Murphy recall Haldane's point that its Bernstein's estimator is negatively biased and go on to suggest novel corrected versions of it. With the identification of more HLA antigens, it is not uncommon to find, in certain populations, that the sample data contain no double blanks; that is, every individual reacts to at least one antigen for a given locus. Gart and Nam give a simple score test of a zero true recessive-gene frequency for such situations. Here we examine the extended test of Stevens as a test of this hypothesis. We find that it is fully efficient for two codominant alleles but that when the number exceeds two its efficiency may be 50% or lower or as high as 100%, depending on the number of alleles and the pattern of gene frequencies. The tests are applied to a set of HLA data.
史蒂文斯提出了一种基于伯恩斯坦估计量的哈迪 - 温伯格定律对ABO血型系统的拟合优度检验。南和加特将此检验扩展到广义的类ABO系统，并将其应用于HLA数据。当隐性基因罕见时，休瑟和墨菲提及霍尔丹的观点，即其伯恩斯坦估计量存在负偏差，并进而提出了新的校正版本。随着更多HLA抗原的鉴定，在某些人群中，样本数据中不存在双空白的情况并不罕见；也就是说，对于给定的位点，每个个体至少对一种抗原产生反应。加特和南针对此类情况给出了一个关于零真实隐性基因频率的简单得分检验。在此，我们将史蒂文斯的扩展检验作为该假设的检验进行研究。我们发现，对于两个共显性等位基因，它是完全有效的，但当等位基因数量超过两个时，其效率可能为50%或更低，也可能高达100%，这取决于等位基因的数量和基因频率模式。这些检验应用于一组HLA数据。
