RNA-Seq Power Calculator – Full User Manual

Purpose

This tool estimates the statistical power or required number of replicates for RNA-Seq experiments using the DESeq2 model.

μ (mean expression): Mean read count per isoform/gene. Can be auto-calculated if total reads and number of isoforms are provided.
Total reads: Optional. If filled together with the number of isoforms, overrides manual μ.
Number of isoforms: Required to auto-calculate μ from total reads.
Dispersion (α): Biological variability. Typically between 0.05 and 0.2 for most genes.
log2 fold change (Δ): Expected expression difference between conditions. Δ = 1 implies a 2-fold change.
Replicates per group: Number of biological replicates in each group. Minimum of 3 required.
FDR: Desired false discovery rate. Used to adjust the per-test significance level.
Number of isoforms (m): Total number of genes/transcripts tested.
Target rank (i): Defines how many top hits you want to discover. Lower rank = stricter test.
Target power: Desired power (e.g. 0.8 = 80%) for planning sample size.

Estimate Power: Calculates the power based on current input values.
Estimate Samples for Target Power: Estimates how many replicates per group are needed to reach the desired power.

Based on the DESeq2 model using a Negative Binomial distribution:

Var(Y) = μ + α·μ²
SE = sqrt((2 / n) × Var / μ²) – Standard error of log2 fold change
Z = |Δ| / SE – Test statistic
αᵢ = (FDR × i) / m – Per-test significance level (Benjamini-Hochberg correction)
Zᵅ = Φ⁻¹(1 − αᵢ) – Critical Z value
Power = Φ(Z − Zᵅ) – Probability of detecting the effect