A Statistical Framework for Family-based Studies in Human GeneticsⅠ

Abstract

The allelic Association between disease genes and marker alleles allows predicting which marker allele is in coupling with the disease. There are novel statistical testing procedures proposed recently. For example, the ransmission/ Disequilibrium Testing (TDT) of Spielman et al. (1993) and its generalizations have been applied to a variety of family-based studies. This approach combines the information of association and linkage to yield a greater power than conventional tests for detecting linkage when the association is present. Although there have been many TDT type procedures proposed, most of them are ad hoc. There is a need for a sensible criteria to evaluate these methods. The purpose of this work is to set up and study a statistical framework that can incorporate procedures, different data sets and other knowledge such as disease information, sampling designs and population information to develop a statistical theory as a useful guideline for future investigators who can design and develop their own statistical procedures to meet their needs in incorporating different environments. In my first talk, I shall introduce the background material, outline our approach and demonstrate how the method works in some special examples including recessive and dominant diseases. The criteria " Locally Most Powerful Testing" is used to evaluate the existing and proposed procedures. We begin in the second talk with extensive simulations and applications to a diabetic data set. We will propose a group of valid LD tests and to evaluate their performances. Finally, we generalize the methods to address variety of scenarios in the family studies.