Seminar Notice
Statistical Laboratory
Iowa State University
DATE AND TIME: Monday, April 25, 2005, 4:00 p.m.
PLACE: 319 Snedecor
SPEAKER: Jason Fine, Department of Statistics and Department of
Biostatistics & Medical Information, University of
Wisconsin-Madison
TITLE: Analysis of Semiparametric Mixture Models with Application to
QTL Analysis
ABSTRACT
In this talk, we propose a semiparametric alternative to traditional
parametric QTL interval mapping. Our model assumes that the log ratio of
the component densities satisfies a linear model, with the baseline density
unspecified. We begin by considering the simple case of single gene model
in backcross, with two component mixtures. We show that a constrained
empirical likelihood has an irregularity when the two densities are equal.
A partial empirical likelihood is proposed, which permits unconstrained
estimation of the parameters and gives consistent and asymptotically normal
estimators. It turns out that the partial empirical likelihood is related
to a conditional likelihood involving additional nuisance parameters. We
establish that the partial likelihood estimator is more efficient than an
estimator with the nuisance parameters known. The practical utility of the
methods is illustrated on a rat study of breast cancer resistance genes.
The approach is then extended to intercross designs and multiple gene
models. The exponential tilt formulation unifies standard parametric
multigene models, including epistasis. Robustness of the
conditional/partial likelihoods to selective genotyping/phenotyping is discussed
and a general resampling technique is described for whole genome scans.
Analysis of a rat experiment of listeria infection will be presented, where
segregation distortion leads to different results compared to standard
parametric QTL analysis.
COFFEE: 3:30 p.m., 104 Snedecor Hall
Jeanette La Grange
Department of Statistics
102 Snedecor
Iowa State University
Ames, IA 50010-1210
515 294-3440 (office)
515 294-4040 (fax)