initRandrotW: Internal function

In phettegger/randRotation: Random Rotation Methods for High Dimensional Data with Batch Structure

Internal function

Description

Internal function

Usage

initRandrotW(Y, X, coef.h, coef.d, weights, cormat, cholCinv, tcholC)

Arguments

Y	a data matrix with features x samples dimensions or a list with elements E, design and weights (see Details). Missing values (NA) are allowed but e.g. lead to NAs for all samples of the respective features in the rotated dataset and should thus be avoided. We highly recommend avoiding missing values by e.g. replacing them by imputation or removing features containing NAs.
X	the design matrix of the experiment with samples x coefficients dimensions. For initBatchRandrot, specify the design matrix without the batch variable. A warning is generated if X[,coef.d] does not have full rank, see Details.
coef.h	single integer or vector of integers specifying the "hypothesis coefficients" (H0 coefficients). coef.h should correspond to the last columns in X (see Details). If available, attr(X, "coef.h") is used, see contrastModel. By default, all coefficients are set as H0 coefficients. If coef.h is set -1, no coefficient is set as H0 coefficient.
coef.d	Determined coefficients. These are all other coefficients that are not hypothesis coefficients (see also initRandrot).
weights	numerical matrix of finite positive weights > 0 (as in weighted least squares regression. Dimensions must be equal to dimensions of Y.
cormat	the sample correlation matrix with samples x samples dimensions. Must be a real symmetric positive-definite square matrix. See Details for usage in initBatchRandrot.
cholCinv	Inverse of the Cholesky factorisation of cormat.
tcholC	Transposed of the Cholesky factorisation of cormat.

Value

An initialised initRandrotW object.

Author(s)

Peter Hettegger

Examples

# For further examples see '?rotateStat' and package vignette.

#set.seed(0)

# Dataframe of phenotype data (sample information)
# We simulate 2 sample classes processed in 3 batches
pdata <- data.frame(phenotype = rep(c("Control", "Cancer"), c(5,5)))
features <- 100

# Matrix with random gene expression data
edata <- matrix(rnorm(features * nrow(pdata)), features)
rownames(edata) <- paste("feature", 1:nrow(edata))

mod1 <- model.matrix(~phenotype, pdata)

### Simulate weights
weights <- matrix(rbeta(features * nrow(pdata), 2, 2)+0.1, features)

# Initialisation of the random rotation class
init1 <- initRandrot(Y = edata, X = mod1, coef.h = 2,
                        weights = weights)
init1

phettegger/randRotation documentation built on April 10, 2023, 7:25 p.m.