Jinks J L, Towey P
Heredity (Edinb). 1976 Aug;37(1):69-81. doi: 10.1038/hdy.1976.66.
A new method, genotype assay, is described for estimating k the number of genes or more strictly the number of effective factors responsible for variation of a continuous kind. The central feature is the determination of the proportion of individuals in the Fn generation of a cross between two pure breeding lines that are heterozygous at, at least, one locus by an assay of their Fn+2 grand progeny families. The observed proportion is then equated to a theoretical expectation which is a function of the number of genes involved. Expectations generalised to cover any generation n for experimental designs in which every Fn individual is assayed by comparing two Fn+2 grand progeny families have been derived for two limiting cases; one in which all genotypic differences are expressed as phenotypic differences and the other where the expression is minimised by imposing the maximum and relational balancing out of the contributions of individual gene loci. Equating the observed proportion of heterozygotes to these expectations therefore, leads to an upper and a lower estimate of k corresponding with these two limiting conditions. The reliability and sensitivity of the estimates depends primarily on n the generation chosen for study, the number of individuals (m) assayed from that generation and the number of individuals (l) raised in each Fn+2 grand progeny family. The two variables m and l being the principal determinants of the variances of the family means set the lower limit to the size of the gene effects that can be detected. The method is illustrated by assays of the F3 and F5 generations of two crosses between conditioned lines of Nicotiana rustica for three characters. The estimates are, without exception, as great as or greater than those obtained by alternative procedures. They show large, consistent increases between the F3 and F5 that cannot be traced to greater sensitivity of the latter generation and hence are presumably genuine.
本文描述了一种新的方法——基因型分析，用于估计k值，即负责连续型变异的基因数量，或者更准确地说是有效因子的数量。其核心特点是通过对Fn+2代大后代家系的分析，确定两个纯系杂交的Fn代中至少在一个位点杂合的个体比例。然后将观察到的比例与一个理论预期相等，该理论预期是所涉及基因数量的函数。对于两种极限情况，已经推导出了适用于任何世代n的预期，在这些实验设计中，通过比较两个Fn+2代大后代家系来分析每个Fn个体；一种情况是所有基因型差异都表现为表型差异，另一种情况是通过对各个基因位点的贡献施加最大程度的相对平衡，使表达最小化。因此，将观察到的杂合子比例与这些预期相等，会得出与这两种极限条件相对应的k的上限和下限估计值。估计值的可靠性和敏感性主要取决于用于研究的世代n、从该世代中分析的个体数量（m）以及每个Fn+2代大后代家系中培育的个体数量（l）。两个变量m和l是家系均值方差的主要决定因素，它们设定了可检测到的基因效应大小的下限。通过对黄花烟草的两个条件品系之间的三个性状的F3代和F5代进行分析来说明该方法。无一例外，这些估计值与通过其他方法获得的估计值一样大或更大。它们在F3代和F5代之间显示出大幅、一致的增加，这不能归因于后一代更高的敏感性，因此大概是真实的。
