Number of independent components proposal function
This function takes the output of 'PriorNormPCA' and returns for a given threshold the number of components to be inferred for subsequent ICA.
proposeNCP(prPCA, thresh = 0.1)
prPCA |
The output object from 'PriorNormPCA'. |
thresh |
Threshold on eigenvalues. |
A list with following components:
X: Normalised data matrix.
x: Normalised data matrix projected onto selected subspace.
pEx: Selected eigenvectors defining subspace for projection.
pCorr: Projected correlation matrix.
ncp: The dimension of the selected subspace(=number of independent components to be inferred with subsequent ICA).
Andrew Teschendorff a.teschendorff@ucl.ac.uk
Hyvaerinen A., Karhunen J., and Oja E.: Independent Component Analysis, John Wiley and Sons, New York, (2001).
Kreil D. and MacKay D. (2003): Reproducibility Assessment of Independent Component Analysis of Expression Ratios from DNA microarrays, Comparative and Functional Genomics *4* (3),300-317.
Liebermeister W. (2002): Linear Modes of gene expression determined by independent component analysis, Bioinformatics *18*, no.1, 51-60.
## The function is currently defined as
function (prPCA, thresh = 0.1)
{
X <- prPCA$X
eigenvals.v <- diag(prPCA$Dx)
Ex <- prPCA$Ex
ntp <- nrow(X)
ndim <- ncol(X)
print("About to find ncp")
p.cpts <- eigenvals.v[eigenvals.v > thresh]
ncp <- length(p.cpts)
pCorr <- diag(eigenvals.v[1:ncp])
pEx <- Ex[, 1:ncp]
x <- matrix(nrow = ntp, ncol = ncp)
for (g in 1:ntp) {
for (c in 1:ncp) {
x[g, c] <- sum(X[g, ] * Ex[, c])/sqrt(diag(pCorr)[c])
}
}
return(list(X = X, x = x, pEx = pEx, pCorr = pCorr, ncp = ncp))
}Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.