Nonparametric inference on mtDNA mismatches

Population biologists use fairly elementary methods to estimate parameters associated with DNA mismatch distributions, which serve as one basis for their formulation of an evolutionary history of the population. This paper presents new statistical methodology that allows better estimation of these quantities and their standard errors. It is shown, in particular, how inference based on a chi-square parametrization of the DNA differences can be replaced with a more robust nonparametric analysis. Mismatches of a sample of DNA sequences yield information on the heterogeneity of the population under study. The mode of the distribution of these differences enables estimation of the expansion times for genetically isolated populations; similarly, the steepness of the frontal wave allows an estimate of the size of the initial population.

Duke Scholars

Author David L. Banks Statistical Science