Regularized Estimation of Large-Scale Gene Association Networks using Graphical Gaussian Models

摘要

We combinine regularized regression methods with the estimation of Graphical Gaussian models. A key issue is the estimation of the matrix of partial correlations for high-dimensional data. Since the (Moore-Penrose) inverse of the sample covariance matrix leads to poor estimates in this scenario, standard methods are inappropriate and adequate regularization techniques are needed. The investigated framework includes various existing regression methods (Lasso, Partial Least Squares) as well as two new approaches based on Ridge Regression and adaptive Lasso, respectively. These methods are extensively compared both qualitatively and quantitatively within a simulation study. In addition, all proposed algorithms are implemented in the R package "parcor11, available from the r repository CRAN. This work is based on a joint project with Juliane SchSfer (University of Basel/ETh Zurich) and Anne-Laure Boulesteix (University of Munich). A full paper-including the application to six diverse real data sets-is available (Kramer, Schafer & Boulesteix, 2009).