Chiasma-based models of multilocus recombination: increased power for exclusion mapping and gene ordering.

Cytological evidence indicates that the number of chiasmata which can occur on any given chromosome arm is limited. In this paper a Monte-Carlo simulation study is used to compare the power of a model of multilocus recombination parameterized in terms of chromosome-specific chiasma distributions with the traditional model parameterized by recombination fractions in adjacent intervals. Two specific gene mapping problems are considered: excluding a test locus from a given chromosome map and ordering a test locus with respect to a fixed map of syntenic marker loci. We show that the chiasma-based models require significantly fewer observations to exclude or order a test locus and that they are quite robust to errors in specifying the underlying true chiasma distribution.