RNA-Seq Power Calculator

RNA-Seq Power Calculator (DESeq2 Model)

Mean expression (μ): Must be > 0.
Total reads (optional):
Number of isoforms (for μ derivation - optional): μ = total reads / isoforms (overrides manual μ if both provided)
Dispersion (α): Must be ≥ 0. Typical: 0.01 (low variance) to >0.5 (high variance). Estimate from pilot data or literature.
log2 fold change (Δ):
Replicates per group (n): Minimum of 3 required.
Target FDR: Must be > 0 and < 1.
Total number of isoforms tested (m): Must be ≥ 1.
Target rank (i): Rank of p-value for setting significance threshold via FDR (e.g., if expecting 1000 discoveries, rank=1000 checks power at that discovery threshold). Must be ≤ m. Lower rank = stricter.
Target power (optional for sample size): Used to suggest sample size (0.1 to 0.999) and evaluate current power.

Methodology & Formulas

Model: DESeq2 (Negative Binomial): Var(Y) = μ + α·μ²
SE of log2FC: SE = sqrt((2 / n) × Var / μ²)
Test statistic: Z = |Δ| / SE
Per-test significance level (α_effective): α_effective = (FDR × i) / m
Power: Φ(Z − Z_α/2) where Z_α/2 = Φ⁻¹(1 − α_effective/2). (Φ is the standard normal CDF)
μ auto-calculation: If total reads and isoforms for μ derivation provided, μ = totalReads / isoforms