Testing differential expression in nonoverlapping gene pairs: a new perspective for the empirical Bayes method

Area of Science:

• Genomics
• Bioinformatics
• Statistical Genetics

Background:

• Current significance testing in microarray gene expression profiling suffers from instability and low statistical power.
• Strong, long-range correlations in gene expression data contribute to these limitations.

Purpose of the Study:

• To introduce a novel methodology for selecting differentially expressed genes using a unique data structure called the delta-sequence.
• To enhance the stability and power of gene expression analysis while eliminating technical noise.

Main Methods:

• Identification of the delta-sequence structure in gene expression data, enabling the use of weakly dependent random variables.
• Development of a new gene selection methodology applied to nonoverlapping gene pairs.
• Integration with a nonparametric empirical Bayes method, leading to performance improvements.

Main Results:

• The proposed method significantly increases the mean number of true discoveries and reduces false discoveries.
• Results demonstrate enhanced stability in significance testing outcomes.
• The methodology is free from log-additive array-specific technical noise.

Conclusions:

• The delta-sequence offers a new paradigm for analyzing biological data, improving gene expression profiling.
• The modified empirical Bayes method shows substantial performance gains.
• This approach provides a robust framework for future methodological research in genomics.