Simulation, power evaluation and sample size recommendation for single-cell RNA-seq.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Sample size determination is critical for high-throughput experiments, particularly for detecting differential expression (DE).
• Existing methods for sample size calculation are limited for single-cell RNA sequencing (scRNA-seq) due to data sparsity and heterogeneity.
• Understudied sample size and power considerations in scRNA-seq DE analysis necessitate new approaches.

Purpose of the Study:

• To introduce POWSC, a novel simulation-based method for power evaluation and sample size recommendations in scRNA-seq DE analysis.
• To address the challenges posed by scRNA-seq data characteristics like sparsity and heterogeneity in DE studies.
• To provide a tool that aids researchers in designing robust scRNA-seq experiments.

Main Methods:

• POWSC utilizes a data simulator to generate realistic scRNA-seq expression data, outperforming existing simulators in capturing key data characteristics.
• A power assessor component evaluates power and sample size relationships through various analyses, including stratified and marginal power.
• The method supports DE detection under different scenarios (phase transition, magnitude tuning) and optimizes sample size versus sequencing depth trade-offs.

Main Results:

• The POWSC data simulator demonstrates superior performance in mimicking real scRNA-seq data properties.
• The power assessor provides comprehensive evaluations, enabling informed decisions on sample size and sequencing depth.
• POWSC facilitates accurate power assessment for differential expression in scRNA-seq data.

Conclusions:

• POWSC offers a robust solution for sample size determination and power analysis in scRNA-seq DE studies.
• The R package provides researchers with a valuable tool to enhance experimental design and statistical power.
• This method contributes to addressing the understudied area of sample size calculation in single-cell genomics.