Although genetic studies have been spectacularly successful in detecting genes responsible for singlelocus Mendelian disorders, a multitude of problems plague genetic studies of complex disorders. Loci of small effect, genetic and ethnic heterogeneity, and missing parental and founder genotypes decrease power and increase rates of false positives. A wide variety of analytic techniques are available, each with its own separate hypotheses and literature. Unfortunately, power for these methods is often tested with normally distributed traits, genetically homogeneous populations, large effect loci, and fully informative pedigrees. Real complex disorders rarely meet these conditions; this may lead to overly optimistic beliefs about the ability to localize genes for complex disorders and subsequent disappointment when putative disease loci fail to replicate in other samples. Correcting for non-normal distributions and mixed ethnicities may increase power and reduce false positives The overall theme of this proposal is robust linkage analysis. In particular, the hypotheses underlying linkage studies and the potential for performing corrections for data that violates these hypotheses will be examined over the course of the mentored career development award. The training will focus on developing an understanding of psychiatric nosology through didactic coursework, patient presentations, and formal and informal lectures. These training goals are designed to provide the psychiatric, genetic, and statistical skills necessary to understand the hypotheses underlying linkage studies of psychiatric disorders and to determine how to measure and adjust phenotypes, genetic marker data, and computation. The research goals of this proposal are to perform a set of simulation studies to develop and evaluate novel methods of analyzing ordinal data and non-normally distributed data, to adjust for mixed and rare ethnicities in linkage methods, and to adapt linkage studies to an association framework. Once developed, these methods will be applied to data from the Collaborative Study of the Genetics of Alcoholism (COGA). Novel, genetically relevant phenotypes will be derived from the COGA data using experience gained from the training component. Additionally, a publicly accessible online archive will be created to freely provide simulated data and software developed during this award.
Showing the most recent 10 out of 12 publications
