glmroast: Multiple Contrast Rotation Tests based on Marginal...

In daniel-gerhard/mcroast: Multiple Contrast Rotation Tests

Description

Adaption of the roast function in package limma to perform a maxT-test for parameters in multiple generalized linear models, controlling FWER for multiple contrasts.

Usage

glmroast(formula, data = NULL, K, family=gaussian(), 
         nrot = 9999, adjusted = TRUE, responsenumber=NULL, ...)

Arguments

formula	A formula object, analogously to lmer, with a matrix of response variables (independent subjects in rows) on the left, and design covariates + random effects on the right hand side.
data	A data.frame object, searching for column names in formula.
K	A contrast matrix (different contrasts in rows), with the same number of columns as number of parameters in the model
family	An object of class family, or a list with family objects for each marginal model.
nrot	Number of Monte Carlo simulation steps
adjusted	logical, if TRUE single-step maxT FWER control is performed; if FALSE the comparison-wise error rate is controlled
responsenumber	An integer vector, as a unique identifier of the response for each marginal model. Default is NULL, being equal to a sequence of integers from 1 to the number of response matrix columns.
...	Additional arguments passed to the glm function.

Details

Most of the code was taken from function roast in package limma (Bioconductor).

Value

A list with

Y	The matrix of response variables
X	The design matrix
statistic	A matrix of test statistics
pvalue	A matrix of (adjusted) p-values
B	A matrix with contrast parameter estimates
mrot	A vector with simulated maxT statistics under H0

Author(s)

Daniel Gerhard

References

Wu, D, Lim, E, Francois Vaillant, F, Asselin-Labat, M-L, Visvader, JE, and Smyth, GK (2010). ROAST: rotation gene set tests for complex microarray experiments. Bioinformatics 26, 2176-2182.

Langsrud, O (2005). Rotation tests. Statistics and Computing 15, 53-60.

Examples

## reproducing the attitude towards science data analysis
## presented in the help pages of function mmm() (package multcomp)
data("mn6.9", package = "TH.data")
glmroast(cbind(y1, y2, y3, y4) ~ group, data=mn6.9, 
         K=diag(2), family=binomial(), nrot=999)

daniel-gerhard/mcroast documentation built on May 14, 2019, 3:38 p.m.