Sample size reassessment for a two-stage design controlling the false discovery rate

Area of Science:

• Genomics
• Biostatistics
• Bioinformatics

Background:

• Accurate sample size calculation is crucial for the statistical power of gene expression microarray and next-generation sequencing RNA-Seq studies.
• Key parameters like the proportion of true null hypotheses and effect size distributions are often unknown during the planning phase, complicating sample size determination.

Purpose of the Study:

• To propose a novel two-stage adaptive design for sample size calculation in high-dimensional genomic studies.
• To address the challenges posed by unknown distributional parameters in microarray and RNA-Seq experiments.
• To ensure adequate statistical power and control the false discovery rate (FDR).

Main Methods:

• A two-stage study design incorporating an adaptive interim analysis.
• Estimation of unknown parameters (proportion of true null hypotheses, effect size distribution) from interim data.
• Dynamic adjustment of the second-stage sample size based on interim estimates to achieve a prespecified overall power.

Main Results:

• The proposed two-stage design effectively estimates unknown quantities from interim data.
• The method successfully determines the second-stage sample size to achieve desired statistical power.
• The procedure maintains control over the false discovery rate (FDR) even with data-dependent sample size adjustments.
• Power is reliably controlled across scenarios, with a minor exception for extremely small first-stage sample sizes.

Conclusions:

• The two-stage adaptive design offers a robust solution for sample size determination in high-dimensional studies with initial uncertainty.
• This approach is particularly valuable when planning gene expression microarray and NGS-RNA-Seq experiments where effect sizes and variability are poorly characterized.
• The method provides a practical tool for researchers to optimize sample size, enhancing the reliability of genomic findings.