Empirical likelihood-based inference for genetic mixture models

The authors show how the genetic effect of a quantitative trait locus can be estimated by a non-parametric empirical likelihood method when the phenotype distributions are completely unspecified. They use an empirical likelihood ratio statistic for testing the genetic effect and obtaining confidence intervals. In addition to studying the asymptotic properties of these procedures, the authors present simulation results and illustrate their approach with a study on breast cancer resistance genes.