RHT.2samp: Two-sample Regularized Hotelling's T-square Test

Description

This function tests if a pathway (or gene set) consists of any protein (or gene) that shows different mean abundance (or expression) between two groups of samples.

Usage

RHT.2samp(path.idx, datX, datY, nsim = 1000, seed = 123)

Arguments

path.idx

This is a LIST. Each element in the list contains the indice of proteins (or genes) for a pathway in the data set.

datX

An N1 by p matrix of protein abundance (or gene expression) from one group of samples. Each row represents one sample and each column represents a protein (or a gene).

datY

An N2 by p matrix of protein abundance (or gene expression) from another group of samples. Each row represents one sample and each column represents a protein (or a gene).

nsim

Number of resamples needed to calculate the p-value. By default, nsim=1000.

seed

A single integer that controls the random number generator in the resampling.

Value

The function returns the p-values for each pathway in the list path.idx.

References

Chen LS, Paul D, Prentice RL and Wang P. (2011) A regularized Hotelling's T-square test for pathway analysis in proteomics studies. Journal of the American Statistical Association, in press.

Examples

Run this code

  ## We simulate a data set X with N=10 samples and p=50 proteins,
  ## and a second data set Y with N=8 sample and the same number of proteins. 
  ## 20% of the data are missing.
  
  
  set.seed(1)
  X <- matrix(rnorm(500),nrow=10)
  X[sample(1:500, 0.2*500)] <- NA
  
  Y <- matrix(rnorm(400),nrow=8)
  Y[sample(1:400, 0.2*400)] <- NA
  
  ## Among the 50 proteins, we randomly assign 2 pathways, with 5 and 12 proteins, respectively.
  path.idx <- list()
  path.idx[[1]] <- 1:5
  path.idx[[2]] <- 13:24
  names(path.idx) <- c("pathway A", "pathway B")
  
  ## The following function tests each pathway to see
  ## if any of the proteins in each pathway shows different 
  ## abundance/expression between data X and Y.
  
  pval <- RHT.2samp(path.idx, datX=X, datY=Y)