Plant Genome IV Conference Plant Genome IV Conference
Statistical properties, such as variance, bias, distribution, and confidence intervals, are critical for interpreting estimated recombination frequencies. For backcross model, estimated recombination frequency is distributed as a binomial variable. However, statistical properties for estimated recombination frequencies using populations other than backcross are largely based on the optimum properties of maximum likelihood estimation. For larger-sample, maximum likelihood estimate is asymptotically normally distributed with mean of the parameter and variance as inverse of the information. In this poster, the way to obtain empirical statistical properties will be described using F2 population as an example. The empirical statistical properties will be compared with the theoretical properties. Estimated variance of recombination frequency may be lower than the Crame-Rao lower bound. Nonparametric confidence intervals for recombination frequency have better properties than normal approximation confidence intervals.