Last data update: 2014.03.03

 R: Generates a covariance matrix.
mPhen.sampleCovarR Documentation

Generates a covariance matrix.

Description

This function can be use to sample covariance matrices. This is useful when simulating data to test Multiphenotype based association strategies. This function lets the user decide on the orthoganality within 'blocks' and between 'blocks' of correlated variables/

Usage

mPhen.sampleCovar(noPhenos,blockSize, orthogAll = c(0.9,0.5),
dirichletScale = 50,  resample = FALSE, 
sd = rgamma(noPhenos,shape=10,rate = 10))

Arguments

noPhenos

The number of phenotypes to simulate

blockSize

The number of phenotypes per covariance block

orthogAll

The orthogonality relationships between and within blocks expressed as a number on the interval (0,1). A number closer to one indicates closer to orthogonality, whereas 0 indicates non-orthogonality. First number is orthogonality between blocks, second is orthogonality within blocks.

dirichletScale

When sampling off diagonal elements of the cholesky decomposition, how much deviation from uniform to allow. Should be a number in interval (0,+Inf). Smaller value leads to greater variation

resample

Whether to randomly shuffle phenotype columns after sampling.

sd

Standard deviation for each phenotype

Value

Simulated covariance matrix

Results


Created & Maintained by Osamu Ogasawara (osamu.ogasawara@gmail.com) and IMSBIO