R Culverhouse1. 1. Department of Internal Medicine, Washington University in St. Louis School of Medicine, St. Louis, MO 63110, USA. rculverh@wustl.edu
Abstract
BACKGROUND/AIMS: Many diseases important to public health are not due solely to a single mutation or environmental insult. Instead, complex interactions among multiple genes and environmental exposures likely play crucial roles in the etiology of diverse phenotypes from schizophrenia to chemotherapy response. The Restricted Partition Method (RPM) was designed to detect qualitative genetic and environmental factors contributing to a quantitative trait, even if the contribution is predominantly presented as an interaction (displaying little or no signal in univariate analyses). Although the RPM was developed with the expectation that trait values would be drawn from normal distributions, the algorithm will function if the quantitative trait values are replaced with 0's or 1's indicating control or case status. The aim of this study is to evaluate the performance of the RPM on case-control data. METHODS: Case-control data simulated for this study and data provided to the Pharmcogenetics Research Network Analysis Workshop 2005 were used to assess power and type I error for the RPM in this setting. RESULTS: For the tested data, the RPM displayed good power and type I error very close to nominal rates. CONCLUSIONS: The RPM is an appropriate method for the analysis of case-control data.
BACKGROUND/AIMS: Many diseases important to public health are not due solely to a single mutation or environmental insult. Instead, complex interactions among multiple genes and environmental exposures likely play crucial roles in the etiology of diverse phenotypes from schizophrenia to chemotherapy response. The Restricted Partition Method (RPM) was designed to detect qualitative genetic and environmental factors contributing to a quantitative trait, even if the contribution is predominantly presented as an interaction (displaying little or no signal in univariate analyses). Although the RPM was developed with the expectation that trait values would be drawn from normal distributions, the algorithm will function if the quantitative trait values are replaced with 0's or 1's indicating control or case status. The aim of this study is to evaluate the performance of the RPM on case-control data. METHODS: Case-control data simulated for this study and data provided to the Pharmcogenetics Research Network Analysis Workshop 2005 were used to assess power and type I error for the RPM in this setting. RESULTS: For the tested data, the RPM displayed good power and type I error very close to nominal rates. CONCLUSIONS: The RPM is an appropriate method for the analysis of case-control data.