bootstrap.build: Perform network inference with bootstrap approach.

In tuxette/internet3D: gamma-package for internet3D

Description

Perform network inference with bootstrap approach.

Usage

bootstrap.build(expr, nboot = 10, nkeep1 = round(0.1 * nrow(expr)^2/2),
  nkeep2 = round(0.5 * nboot), parallel = FALSE, ncores = 10,
  seeds = NULL, ...)

Arguments

expr	a n by d data matrix
nboot	number of bootstrap samples
nkeep1	number of edges to keep from each infered network
nkeep2	minimum number of times an edge has to be predicted to be kept in the final network
parallel	should the function be run in parallel (unix-like systems only)
ncores	number of cores for parallel executions
seeds	vector of nboot seeds to control randomness in network inference based on bootstraps
...	options passed to the function set.options

Value

object with the following entries

• edges final infered network

• mse total OOB mean square error

• adjacency total adjacency matrix combined from the nboot infered network

• stars StaRs criterion

• stability stability criterion

Examples

data(cancer)
# no constraint
res <- bootstrap.build(expr, nboot=3, nkeep1=10, nkeep2=1)
# constraints
res <- bootstrap.build(expr, nboot=3, nkeep1=10, nkeep2=1,
                       first.list=matrix(c("AMFR","BB_S4","BECNI","BTG3"),
                                         ncol=2, byrow=TRUE),
                       second.list=matrix(c("AMFR","E2F3"), ncol=2),
                       mu=1)
print(res)

tuxette/internet3D documentation built on May 8, 2019, 11:59 p.m.