Pooled DNA for linkage analysis: practical and statistical considerations

Abstract
In an F2 population, the alleles at two loci with a recombination fraction r < 0.5 are in linkage disequilibrium. If r is small, then a pool of DNA from k diploid individuals that are fixed at one locus has a relatively high probability (P = (1 − r)2k) of containing only the coupled allele at the second locus. Based on this principle, several methods have been described to detect linkage (using one or two pools) or to estimate r (using a group of n pools). This report compares maximum likelihood and approximate estimators of r for use in pooled-DNA analysis and illustrates the use of this analysis for dominant markers. Expected values and expected mean squares for estimators of r were computed for varying levels of r, k, and n. Both estimators were biased, but the bias and variability were slightly smaller for the maximum-likelihood estimator. Bias was not severe except when k was large relative to r and (or) n. Methods for optimizing k are discussed relative to several criteria: minimizing variance, minimizing bias, minimizing the probability that linkage cannot be detected, and minimizing the number of samples that must be screened.Key words: pooled DNA, linkage analysis, molecular markers, RAPD, genetic mapping.