Power to detect linkage with heterogeneity in samples of small nuclear families

Abstract

Computer simulation methods were used to investigate the power of genetically homogeneous or heterogeneous samples of nuclear families to detect linkage of a rare dominant disease allele to flanking DNA markers (threepoint analysis, admixture text). Phase was assumed to be unknown (no grandparents available), and unaffected siblings were not considered. A sample of 95 families with an ill parent and two ill offspring, or 45 families with three ill offspring, demonstrated 90% power to detect a lod score of 3.0 when 50% of families were assumed to be segregating for a disease allele located midway between two DNA markers (PIC = .70) that were .05 M apart. When the proportion of linked families (α) = .25, 90% power required 380 and 160 families, respectively. For α < .25, sample size requirements become prohibitive. Issues are reviewed concerning the use of the admixture test in the case of more complex disease models. Screening of the genome with adequate sample sizes for low values of α is likely to require multiple large collaborative efforts.