Estimating effect sizes of differentially expressed genes for power and sample-size assessments in microarray...

Area of Science:

• Genomics
• Bioinformatics
• Statistical Genetics

Background:

• Microarray screening requires accurate power and sample size assessment for reliable gene expression analysis.
• Overestimation of effect sizes due to random variation is a challenge in conventional methods.
• Differential gene expression analysis necessitates robust estimation of effect sizes for accurate power calculations.

Purpose of the Study:

• To propose a simple estimation method for gene effect sizes using hierarchical mixture models.
• To improve the assessment of power and false discovery rate (FDR) in microarray gene screening.
• To introduce a novel 'partial power' index for practical gene selection in experiments.

Main Methods:

• Utilized hierarchical mixture models with a nonparametric prior distribution.
• Employed empirical Bayes estimates for effect sizes of differentially expressed genes.
• Developed a 'partial power' index for evaluating the selection of top genes.

Main Results:

• The proposed method accommodates random variation and diverse effect sizes, distinguishing between differential and non-differential genes.
• Simultaneous estimation of power and FDR is achievable for effective gene screening.
• The 'partial power' index offers a practical solution for power limitations in microarray experiments.

Conclusions:

• The new method provides a more accurate estimation of effect sizes, leading to improved power and FDR assessment.
• This approach enhances the reliability of gene screening in microarray studies, particularly in cancer research.
• The 'partial power' index presents a valuable tool for researchers facing power constraints in gene expression studies.