| Two procedures were developed for identifying more heterozygous sires as family sires for QTL mapping under half-sib designs. First, a new index was proposed to measure an individual's heterozygosity of genes influencing a trait. It was shown that sire's heterozygosity index completely determines the across-family variation of within-family additive genetic variance and is equal to the upper limit of variance among marker genotypes. A least squares method is presented to estimate sire's heterozygosity using progeny phenotypes; more heterozygous sires were subsequently selected using simulated data. As expected, selection response measured by the true heterozygosity ratio of selected sires to all sires depended mainly on estimation accuracy. The heterozygosity ratio changed slightly when using daughters' phenotypes, but greatly increased when using progeny-tested sons, with its magnitude depending on the genetic model of the trait, family size and structure, and selection intensity.; Second, a two-stage procedure for choosing sires and chromosomes, in contrast to the conventional one-step procedure, was proposed and evaluated. Based on analyzing small-scale marker data accumulated at stage 1, sires and chromosomes are selected for stage 2. Results indicate that the two-stage design is more powerful with large noncentrality parameter, large QTL and family size, and large variation in QTL number and sizes among chromosomes.; A procedure was also developed to choose optimal markers that maximize experimental power for a uniformly distributed QTL. Results indicate that optimal marker choices were affected considerably by DNA sample preparation cost, while little influence was found for the amount of financial resources. Lower sample preparation cost favored genotyping fewer linked markers, and allelic frequencies were more influential factors than marker spacing.; Finally, a maximum likelihood method to construct parental haplotypes using progeny genotypes is presented. Exact error rates and family size choices in haplotype construction were evaluated mathematically and numerically. Results suggest that a difference of one between progeny receiving parental and recombinant gametes is sufficient for constructing sire phase of two tightly linked markers; differences of 2 are required with two markers 30 centiMorgans apart. Sample size needed for a given power rate decreased considerably as the number of marker alleles increased. |
