interimTrialSimulation: Simulation of Several Survival Studies With the Same...

In survGenesInterim: Simulation of Survival Studies with Microarray Data

Description

interimTrialSimulation performs several simulations of survival studies based on the given parameters. For each survial study simulation the patient data (arrival times and survival times) and the gene expression level data are newly generated.

Usage

  interimTrialSimulation(fc, Sigma.1, Sigma.2, N, l.1.tick,
    l.2, lambda, M.1, M.2, alpha, powerThreshold,
    adjustment, numSimRuns, parallel = TRUE)

Arguments

fc	the vector of foldchanges between the two groups
Sigma.1	the covariance matrix describing the correlation between the genes in group one
Sigma.2	the covariance matrix describing the correlation between the genes in group two. If this is NULL, the homoscedastic case is assumed and Sigma.2 is set to Sigma.1.
N	the sample size, i.e. the number of patients
l.1.tick	the (anticipated) lenth of the recruitment period
l.2	the length of the follow-up period to be simulated
lambda	the mean survival time of the patients
M.1	the number of analyses during the recruitment period to be simulated
M.2	the number of analyses during the follow-up period to be simulated
alpha	the error level at which the FDR is to be controlled
powerThreshold	the study is stopped as soon as the estimated power rate exceeds this threshold
adjustment	the method to use for the p-value adjustment to account for multiple testing. In c("holm", "hochberg", "hommel", "bonferroni", "BH", "BY", "fdr", "none")
numSimRuns	the number of survival studies to be simulated
parallel	boolean value specifying whether to use the package 'multicore' for a parallel execution of the simulation loop

Value

resultTable

the only real result. This table holds the mean error and power rates from all the simulation runs. The first row additionally shows for each (interim) analysis the fraction of simulated survival studies that were stopped then.

The other values are copied versions of the corresponding input parameters. They are included into the result for convenience only.

Author(s)

Andreas Leha [e-mail]

See Also

calls generatePatientData and generateExpressionData to generate the data and interimTrial to simulate a single survival study

plotinterimFDR, plotinterimAPR, and plotinterimStops to visualize the results

Examples

## parameters controlling the gene level expression
fc <- c(rep(0, 500), ceiling(rnorm(500, 0, 1)-0.5))
Sigma.1 <- diag(1000)
Sigma.2 <- diag(1000)

## parameters controlling the patient data
N <- 50
l.1.tick <- 60
l.2 <- 60
lambda <- 60

## parameters controlling the study design
M.1 <- 2
M.2 <- 2
alpha <- 0.05
powerThreshold <- 0.8
adjustment <- "BH"

## the number of studies to simulate
numSimRuns <- 2

## Not run: result <- interimTrialSimulation(fc, Sigma.1, Sigma.2,
                                          N, l.1.tick, l.2, lambda,
                                          M.1, M.2, alpha, powerThreshold, adjustment,
                                          numSimRuns, parallel=FALSE)
## End(Not run)
## Not run: result$resultTable

survGenesInterim documentation built on May 2, 2019, 5:22 p.m.