Shrinkage Discriminant Analysis and CAT Score Variable Selection

This package provides an efficient framework for high-dimensional linear and diagonal discriminant analysis with variable selection. The classifier is trained using James-Stein-type shrinkage estimators and predictor variables are ranked using CAT scores (correlation-adjusted t-scores). Variable selection error is controlled using false non-discovery rates or higher criticism scores.

Current Version: 1.2.1

Authors: Miika Ahdesmäki, Verena Zuber, and Korbinian Strimmer.

Documentation and Installation:

• Manual (pdf file) and release history.
• Download of sda version 1.2.1 source package.
• Archive of previous versions of sda.
• Licensed under the GNU GPL version 3 (or any later version).

Additional Material:

• Example R script for analyzing Khan et al. (2001) data.
• Example R script for feature selection using Singh et al. (2002) data.

Relevant Publication:

• Ahdesmäki, M., and K. Strimmer. 2010. Feature selection in omics prediction problems using cat scores and false non-discovery rate control. Ann. Appl. Stat. 4: 503-519. (reprint, arXiv:0903.2003)
• Zuber, V., and K. Strimmer. 2009. Gene ranking and biomarker discovery under correlation. Bioinformatics 25: 2700-2707. (reprint, arXiv:0902.0751)

Back to software page.