Wong F L, Rotter J I
Am J Hum Genet. 1984 Nov;36(6):1279-97.
Segregation analysis, employing nuclear families, is the most frequently used method to evaluate the mode of inheritance of a trait. To our knowledge, there exists no tabular information regarding the sample sizes required of individuals and families needed to perform a significance test of a specific segregation ratio for a predetermined power and significance level. To fill this gap, we have developed sample-size tables based on the asymptotic variance of the maximum likelihood estimate of the segregation ratio and on the normal approximation for two-sided hypothesis testing. Assuming homogeneous sibship size, minimum sample sizes were determined for testing the null hypothesis for the segregation ratio of 1/4 or 1/2 vs. alternative values of .05-.80, for the significance level of .05 and power of .8, for ascertainment probabilities of nearly 0 to 1.0, and sibship sizes 2-7. The results of these calculations indicate a complex interaction of the null and the alternate hypotheses, ascertainment probability, and sibship size in determining the sample size required for simple segregation analysis. The accompanying tables should aid in the appropriate design and cost assessment of future genetic epidemiologic studies.
采用核心家庭的分离分析是评估性状遗传模式最常用的方法。据我们所知，对于在预定的检验效能和显著性水平下对特定分离比进行显著性检验所需的个体和家庭样本量，尚无表格信息。为填补这一空白，我们基于分离比最大似然估计的渐近方差以及双侧假设检验的正态近似，编制了样本量表格。假设同胞大小相同，针对检验分离比为1/4或1/2的原假设与0.05 - 0.80的备择值，显著性水平为0.05、检验效能为0.8，确定概率接近0至1.0，以及同胞大小为2 - 7的情况，确定了最小样本量。这些计算结果表明，在确定简单分离分析所需样本量时，原假设和备择假设、确定概率以及同胞大小之间存在复杂的相互作用。随附的表格应有助于未来遗传流行病学研究的合理设计和成本评估。
