A statistical procedure for detecting highly correlated genes with a pre-specified candidate gene in microarray analysis

被引：2

作者：

Ding, Aidong Adam ^{[1
]}

Lin, Jennifer ^{[3
]}

Niu, Tianhua ^{[2
]}

机构：

[1] NE Univ, Dept Math, Boston, MA 02115 USA

[2] Harvard Univ, Sch Publ Hlth, Dept Epidemiol, Boston, MA 02115 USA

[3] Harvard Univ, Sch Publ Hlth, Brigham & Womens Hosp, Dept Med, Boston, MA 02115 USA

来源：

COMMUNICATIONS IN STATISTICS-THEORY AND METHODS | 2008年 / 37卷 / 18期

基金：

美国国家卫生研究院;

关键词：

array-normal-scores; correlation coefficients; false discovery rate; gene selection; multiple testing;

D O I：

10.1080/03610920801923876

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

We propose a statistically valid two-stage procedure for selecting genes with expression levels correlated with that of a 'seed' gene in microarray experiments: Stage I: perform array-normal-scores (ANS) transformation of the raw microarray data and calculate the Pearson correlation coefficients using ANS-transformed data, and Stage II: calculate a resampling-based BH (rsBH) least significant-false discovery rate (LS-FDR), named LS-FDRANSrsBH to select most correlated genes based on a given LS-FDR threshold. By using simulated data sets, we show that the proposed ANS transformation is needed even after the usual data normalization process. The ANS transformation improves the detection of correlated genes, particularly the negatively correlated genes. We demonstrated the proposed procedure by searching for genes involved in insulin resistance and type 2 diabetes mellitus on a dataset the from Human Gene Expression Index (HugeIndex) database.

引用

页码：2991 / 3007

页数：17