Skip to main content

Table 1 Summary of variance shrinkage used by several procedures

From: A benchmark for statistical microarray data analysis that preserves actual biological and technical variance

Method Equation Shrinkage level Offset Estimation
Regularized t-test
V = S2
Window Estimator
SAM d-statistic (Tusher) a = 1
b = 1
V = S S0 = arg min(CV(MAD(d a )))
Fudge factor minimizing CV
Moderated-t (Limma)
V = S2

ψ, ψ' = digamma, trigamma functions
b = 1 - a
V = S2
  1. The general formulation V = a V0 + b Vg illustrates that the variability estimator is estimated from two terms, respectively the background variability V0 and individual variability Vg.