Strategies for characterizing highly polymorphic markers in human gene mapping.

Abstract

Before new markers are thoroughly characterized, they are usually screened for high polymorphism on the basis of a small panel of individuals. Four commonly used screening strategies are compared in terms of their power to correctly classify a marker as having heterozygosity of 70% or higher. A small number of typed individuals (10, say) are shown to provide good discrimination power between low- and high-heterozygosity markers when the markers have a small number of alleles. Characterizing markers in more detail requires larger sample sizes (e.g., at least 80-100 individuals) if there is to be a high probability of detecting most or all alleles. For linkage analyses involving highly polymorphic markers, the practice of arbitrarily assuming equal gene frequencies can cause serious trouble. In the presence of untyped individuals, when gene frequencies are unequal but are assumed to be equal in the analysis, recombination-fraction estimates tend to be badly biased, leading to strong false-positive evidence for linkage.