Weeks D E, Lange K
Department of Human Genetics, University of Pittsburgh, PA 15261.
Am J Hum Genet. 1992 Apr;50(4):859-68.
The affected-pedigree-member (APM) method of linkage analysis is designed to detect departures from independent segregation of disease and marker phenotypes. The underlying statistic of the APM method operates on the identity-by-state relations implied by the marker phenotypes of the affected within a pedigree. Here we generalize the APM statistic to multiple linked markers. This generalization relies on recursive computation of two-locus kinship coefficients by an algorithm of Thompson. The distributional properties of the extended APM statistic are investigated theoretically and by simulation in the context of one real and one artificial data set. In both examples, the multilocus statistic tends to reject, more strongly than the single-locus statistics do, the null hypothesis of independent segregation between the disease locus and the marker loci.
连锁分析中的患病家系成员(APM)方法旨在检测疾病表型与标记表型独立分离的偏离情况。APM方法的基础统计量作用于家系中患病个体标记表型所隐含的状态一致性关系。在此,我们将APM统计量推广到多个连锁标记。这种推广依赖于通过汤普森算法对两位点亲缘系数进行递归计算。在一个真实数据集和一个人工数据集的背景下,从理论和模拟两方面研究了扩展APM统计量的分布特性。在这两个例子中,多位点统计量比单位点统计量更倾向于拒绝疾病位点与标记位点之间独立分离的零假设。