Linkage Strategies for Genetically Complex Traits.
II. The Power of Affected Relative Pairs
N. Risch
American Journal of Human Genetics, 46, 229-241 (Feb 1990)
Abstract
The power to detect disease-susceptibility loci through linkage analysis using
pairs of affected relatives and affected-unaffected pairs is examined. Allelic identity by
descent (ibd) for a completely polymorphic marker for sibling, uncle-nephew,
grandparent-grandchild, half-sib, and first-cousin pairs is considered. Affected-unaffected
pairs generally represent a poor strategy. For single-locus models, ibd depends on lambda
R, the risk ratio for type R relatives compared with population prevalence, and the
recombination fraction theta. The ibd for grandparent-grandchild pairs is least affected by
recombination, followed by sibs, half-sib, uncle-nephew, and first-cousin pairs. For
diseases with large lambda values and for small theta values, distant relatives offer greater
power. For larger theta values, grandparent-grandchild pairs are best; for small lambda
values, sibs are best. Additive and multiplicative multilocus models are considered. For the
multiplicative model, the same formulas as in the single-locus model apply, except that
lambda iR (for the ith contributing locus) is substituted for lambda R. For the additive
model, the deviation from null expectation for ibd is divided among all contributing loci.
Compared with the multiplicative model, for an additive model there is usually greater
advantage in distant relationships. Multipoint analysis using linked marker loci for affected
relative pairs is described. Simultaneous use of multiple markers diminishes the effect of
recombination and allows for localization of the disease-susceptibility locus.
Allele-sharing linkage analysis