Empirical power of three preliminary methods for ordering loci.

Overall, the multidimensional scaling method was less powerful than either seriation or the product of recombination frequencies, and the latter two methods were approximately equally powerful.

Abstract

We empirically estimated the power of three methods for ordering loci within a known linkage group. Estimates of pairwise recombination fractions and correlation coefficients were obtained from data on 50 replicates of 50 and 100 pedigrees by using a likelihood method (LIPED; Ott 1974) and the sib-pair test. Locus order then was determined using seriation, multidimensional scaling, and the product of recombination frequencies. Overall, the multidimensional scaling method was less powerful than either seriation or the product of recombination frequencies. The latter two methods were approximately equally powerful. As expected, the power of the sib-pair test was less than half that of the likelihood method.