Multifactor Dimensionality Reduction (MDR) is a method of detecting genetic interactions by exhaustively searching multi-locus combinations (Motsinger-Reif and others 2008; Ritchie and others 2001). In MDR, k (e.g., k=3) factors and their possible multifactor classes are represented in k-dimensional space. Each multifactor class in the space is then labeled ‘high risk’ if the cases-to-controls ratio meets or exceeds some threshold, or as ‘low risk’ if that threshold is not exceeded, thus reducing the k-dimensional space to one dimension with two levels (‘low risk’ and ‘high risk’) (Moore 2003). The best k-locus model is then selected, the model is evaluated against the testing group, and testing accuracy is calculated. The PA is then calculated for the testing set (Motsinger-Reif and others 2008). Pedigree-based generalized MDR (PGMDR), a new generalized MDR for pedigree data, is a non-parametric method based on the score of the generalized linear model, which permits adjustment for covariates and handling of both dichotomous and quantitative phenotypes (Lou and others 2008). A key advantage of PGMDR is that the method can handle different pedigree structures and sizes simultaneously in
