About this Journal Submit a Manuscript Table of Contents
Computational and Mathematical Methods in Medicine
Volume 2013 (2013), Article ID 860673, 11 pages
http://dx.doi.org/10.1155/2013/860673
Research Article

A Robust Rerank Approach for Feature Selection and Its Application to Pooling-Based GWA Studies

1Institute of Statistical Science, Academia Sinica, Taipei 11529, Taiwan
2Institute of Epidemiology and Preventive Medicine, National Taiwan University, Taipei 10055, Taiwan
3Research Center of Genes, Environment and Human Health, National Taiwan University, Taipei 10055, Taiwan

Received 12 January 2013; Accepted 8 March 2013

Academic Editor: Shinto Eguchi

Copyright © 2013 Jia-Rou Liu et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Large- -small- datasets are commonly encountered in modern biomedical studies. To detect the difference between two groups, conventional methods would fail to apply due to the instability in estimating variances in -test and a high proportion of tied values in AUC (area under the receiver operating characteristic curve) estimates. The significance analysis of microarrays (SAM) may also not be satisfactory, since its performance is sensitive to the tuning parameter, and its selection is not straightforward. In this work, we propose a robust rerank approach to overcome the above-mentioned diffculties. In particular, we obtain a rank-based statistic for each feature based on the concept of “rank-over-variable.” Techniques of “random subset” and “rerank” are then iteratively applied to rank features, and the leading features will be selected for further studies. The proposed re-rank approach is especially applicable for large- -small- datasets. Moreover, it is insensitive to the selection of tuning parameters, which is an appealing property for practical implementation. Simulation studies and real data analysis of pooling-based genome wide association (GWA) studies demonstrate the usefulness of our method.