BioMark (version 0.4.5)

aux.biom: Auxiliary functions in the biomarker package

Description

These functions return coefficient sizes for a variety of modelling methods. Not to be called directly by the user - use function get.biom for that.

Usage

pcr.coef(X, Y, ncomp, scale.p, ...)
pcr.stab(X, Y, ncomp, scale.p, segments = NULL, variables = NULL, ...)
pls.coef(X, Y, ncomp, scale.p, ...)
pls.stab(X, Y, ncomp, scale.p, segments = NULL, variables = NULL, ...)
vip.coef(X, Y, ncomp, scale.p, ...)
vip.stab(X, Y, ncomp, scale.p, segments = NULL, variables = NULL, ...)
lasso.coef(X, Y, scale.p, lasso.opt = biom.options()$lasso,...)
lasso.stab(X, Y, scale.p, segments = NULL, variables = NULL, ...)
shrinkt.coef(X, Y, scale.p, ...)
shrinkt.stab(X, Y, scale.p, segments = NULL, variables = NULL, ...)
studentt.coef(X, Y, scale.p, ...)
studentt.stab(X, Y, scale.p, segments = NULL, variables = NULL, ...)
pval.pcr(X, Y, ncomp, scale.p, npermut)
pval.plsvip(X, Y, ncomp, scale.p, npermut, smethod)

Arguments

X

Data matrix. Usually the number of columns (variables) is (much) larger than the number of rows (samples).

Y

Class indication. Either a factor, or a numeric vector.

ncomp

Number of latent variables to use in PCR and PLS (VIP) modelling. In function get.biom this may be a vector; in all other functions it should be one number. Default: 2.

scale.p

Scaling. This is performed individually in every crossvalidation iteration, and can have a profound effect on the results. Default: "none". Other possible choices: "auto" for autoscaling, "pareto" for pareto scaling, "log" and "sqrt" for log and square root scaling, respectively.

segments

matrix where each column indicates a set of samples to be left out of the analysis.

variables

indices of variables to be used in the analysis.

lasso.opt

optional arguments to the glmnet function, in the form of a list.

...

Further arguments for modelling functions. Often used to catch unused arguments.

npermut

Number of permutations to use in the calculation of the p values.

smethod

Either "both", "pls", or "vip" - indicates what coefficients to convert to p values. Both are derived from PLS models so it is much more efficient to calculate them together.

Value

coef return t-statistics or model coefficients for all variables. The functions ending in stab return these statistics in a matrix, one column per segment. The functions starting with pval convert model coefficients or VIP statistics into p values, using permutation resampling.

See Also

get.biom, glmnet, scalefun