MetSizeR (version 2.0.0)

metsize: Metsize

Description

Function to estimate sample size required to achieve a desired statistical power.

Usage

metsize(
  pilot = pilot,
  n1 = 4,
  n2 = 4,
  p = 200,
  prop = 0.25,
  covars = covars,
  ncovar = 0,
  model = "PPCA",
  plot.prop = FALSE,
  target.fdr = 0.05,
  Targeted = FALSE
)

Arguments

pilot

A matrix containing pilot data, if available.

n1

Sample group 1 minimum sample size to consider.

n2

Sample group 2 minimum sample size to consider.

p

Number of bins in the data.

prop

Expected proportion of significant bins.

covars

Matrix of covariates, if available. Categorical covariates should be included as X-1 dummy variables for a covariate with X levels.

ncovar

The number of covariates present. For Z numeric covariates and Y categorical covariates with X total levels, this should be Z + X - Y.

model

The chosen model.

plot.prop

TRUE/FALSE should sample size be calculated for varying proportions of significant bins?

target.fdr

The desired false discovery rate.

Targeted

TRUE/FALSE is the experiment intended as targeted analysis?

Value

List containing the estimated sample size with the per-group sample size breakdown, results of FDR calculations for each sample size tested, results for varying the proportion of significant metabolites, the number of bins input, the number of covariates input, the model type used, the input minimum sample sizes for group 1 and group 2, and the desired FDR.

References

G. Nyamundanda, I. C. Gormley, Y. Fan, W. M. Gallagher and L. Brennan, BMC Bioinformatics, 2013, 14, 338.

Examples

# NOT RUN {
metsize(pilot = NULL, n1 = 4, n2 = 4, p = 200, prop = 0.2, covars = NULL,
    ncovar = 0, model = "PPCA", plot.prop = FALSE, target.fdr = 0.10,
    Targeted = FALSE)
metsize(pilot = NULL, n1 = 6, n2 = 4, p = 100, prop = 0.1, covars = NULL,
    ncovar = 2, model = "PPCCA", plot.prop = FALSE, target.fdr = 0.05,
    Targeted = TRUE)
# }