Experimental design and error detection for polyploid radiation hybrid mapping.

Abstract

In this paper we consider issues of experimental design and error detection and correction for polyploid radiation hybrid mapping. Using analytic methods and computer simulation, we first consider the combinations of fragment retention rate, ploidy, and marker spacing that provide the best chance to order markers. We find that in general, combinations of ploidy and chromosome-specific retention rates that lead to a per-hybrid retention rate of approximately 50% result in the greatest power to order markers. We also find that analyzing polyploid radiation hybrids as if they were haploid does not compromise the ability to order markers but does result in less accurate intermarker distance estimates. Second, we examine the effect of typing errors on two-locus information, ability to order multiple loci, and estimation of intermarker distances and total map length. Even low levels of error result in large losses of information about breakage probabilities, markedly reduce ability to order loci, and inflate estimates of intermarker distances and total map length. We compare the ordering accuracy that results from duplicate typing of hybrids to that of single typing twice as many hybrids and find that duplicate typing results in a higher probability of identifying the true order as one of the best orders, but that single typing of twice as many hybrids results in stronger support for the true order. For low error rates, framework maps constructed from the larger single-typed panels are only slightly less likely to be correct and include substantially more markers than the smaller double-typed panels. Third, we develop a method to calculate the distribution of the number of obligate chromosome breaks for a polyploid radiation hybrid under a given locus order and discuss how this method may be used to identify hybrids with suspiciously large numbers of chromosome breaks.