cv.glmSparseNet: Calculate cross validating GLM model with network-based...

In glmSparseNet: Network Centrality Metrics for Elastic-Net Regularized Models

Description

network parameter accepts:

Usage

cv.glmSparseNet(
  xdata,
  ydata,
  network,
  network.options = networkOptions(),
  experiment.name = NULL,
  ...
)

Arguments

xdata	input data, can be a matrix or MultiAssayExperiment
ydata	response data compatible with glmnet
network	type of network, see below
network.options	options to calculate network
experiment.name	Name of experiment in MultiAssayExperiment
...	parameters that cv.glmnet accepts

Details

* string to calculate network based on data (correlation, covariance) * matrix representing the network * vector with already calculated penalty weights (can also be used directly glmnet)

Value

an object just as cv.glmnet

Examples

## Not run: 
    # Gaussian model
    xdata <- matrix(rnorm(500), ncol = 5)
    cv.glmSparseNet(xdata, rnorm(nrow(xdata)), 'correlation',
                    family = 'gaussian')
    cv.glmSparseNet(xdata, rnorm(nrow(xdata)), 'covariance',
                    family = 'gaussian')

## End(Not run)

#
#
# Using MultiAssayExperiment with survival model


#
# load data
xdata <- MultiAssayExperiment::miniACC

#
# build valid data with days of last follow up or to event
event.ix <- which(!is.na(xdata$days_to_death))
cens.ix <- which(!is.na(xdata$days_to_last_followup))
xdata$surv_event_time <- array(NA, nrow(colData(xdata)))
xdata$surv_event_time[event.ix] <- xdata$days_to_death[event.ix]
xdata$surv_event_time[cens.ix] <- xdata$days_to_last_followup[cens.ix]

#
# Keep only valid individuals
valid.ix <- as.vector(!is.na(xdata$surv_event_time) &
                      !is.na(xdata$vital_status) &
                      xdata$surv_event_time > 0)
xdata.valid <- xdata[, rownames(colData(xdata))[valid.ix]]
ydata.valid <- colData(xdata.valid)[,c('surv_event_time', 'vital_status')]
colnames(ydata.valid) <- c('time', 'status')

#
cv.glmSparseNet(xdata.valid,
                ydata.valid,
                nfolds          = 5,
                family          = 'cox',
                network         = 'correlation',
                experiment.name = 'RNASeq2GeneNorm')

glmSparseNet documentation built on April 14, 2021, 6 p.m.