snpRFcv: Random Forest Cross-Valdidation for feature selection

Description

This function shows the cross-validated prediction performance of models with sequentially reduced number of predictors (ranked by variable importance) via a nested cross-validation procedure.

Usage

snpRFcv(trainx.autosome=NULL,trainx.xchrom=NULL,trainx.covar=NULL, trainy, 
        cv.fold=5, scale="log", step=0.5, 
        mtry=function(p) max(1, floor(sqrt(p))), recursive=FALSE, ...)

Arguments

trainx.autosome

A matrix of autosomal markers with each column corresponding to a SNP coded as count of a particular allele (i.e. 0,1 or 2), and each row corresponding to a sample/individual.

trainx.xchrom

A matrix of X chromosome markers, each marker coded as two adjacent columns, alleles of a marker are coded as 0 or 1 for carrying a particular allele. Although males only have one X-chromosome, their markers are coded as

trainx.covar

A matrix of covariates, each column being a different covariate, and each row, a sample/individual.

trainy

vector of response, must be a factor and have length equal to the number of rows in trainx.*

cv.fold

number of folds in the cross-validation

scale

if "log", reduce a fixed proportion (step) of variables at each step, otherwise reduce step variables at a time

step

if log=TRUE, the fraction of variables to remove at each step, else remove this many variables at a time

mtry

a function of number of remaining predictor variables to use as the mtry parameter in the snpRF call

recursive

whether variable importance is (re-)assessed at each step of variable reduction

...

other arguments passed on to snpRF

Value

A list with the following components: list(n.var=n.var, error.cv=error.cv, predicted=cv.pred)
n.varvector of number of variables used at each step
error.cvcorresponding vector of error rates or MSEs at each step
predictedlist of n.var components, each containing the predicted values from the cross-validation

References

Svetnik, V., Liaw, A., Tong, C. and Wang, T., ``Application of Breiman's Random Forest to Modeling Structure-Activity Relationships of Pharmaceutical Molecules'', MCS 2004, Roli, F. and Windeatt, T. (Eds.) pp. 334-343.

Examples

Run this code

set.seed(647)
data(snpRFexample)
result <- snpRFcv(trainx.autosome=autosome.snps,trainx.xchrom=xchrom.snps,
                  trainx.covar=covariates, trainy=phenotype)
with(result, plot(n.var, error.cv, log="x", type="o", lwd=2))

## The following can take a while to run, so if you really want to try
## it, copy and paste the code into R.

result <- replicate(5,snpRFcv(trainx.autosome=autosome.snps,
                              trainx.xchrom=xchrom.snps,
                              trainx.covar=covariates, trainy=phenotype), 
		    simplify=FALSE)
error.cv <- sapply(result, "[[", "error.cv")
matplot(result[[1]]$n.var, cbind(rowMeans(error.cv), error.cv), type="l",
        lwd=c(2, rep(1, ncol(error.cv))), col=1, lty=1, log="x",
        xlab="Number of variables", ylab="CV Error")