globalNullModel: Generate global null model p-value

In mil2041/netboxr: netboxr

Generate global null model p-value

Description

Randomly select the same number of nodes in the largest connected component of netbox result as a new gene candidate list and repeat multiple times to produce a distribution of node size and edge numbers. This distribution will be used to produce global p-value of netbox result based on the node size or edge numbers of largest component in the final network result.

Usage

globalNullModel(
  netboxGraph,
  networkGraph,
  directed,
  iterations = 30,
  numOfGenes = NULL,
  pValueAdj = "BH",
  pValueCutoff = 0.05
)

Arguments

netboxGraph	igraph network graph object. This igraph object contains NetBox algorithm identified network from geneConnector function
networkGraph	igraph network graph object. This igraph object contains curated network information
directed	boolean value indicating whether the input network is directed or undirected (default = FALSE)
iterations	numeric value for number of iterations
numOfGenes	numeric value for number of genes mapped in the initial network
pValueAdj	string for p-value correction method c("BH", "Bonferroni") as described in the details section (default = "BH")
pValueCutoff	numeric value of p-value cutoff for linker nodes (default = 0.05)

Details

P-value correction methods include the Bonferroni correction ("bonferroni") or Benjamini & Hochberg ("BH").

Value

a list of returned results

• globalNull: data frame of global randomization results

• globalNodesResult: data frame of global null tested results based on nodes

• globalEdgesResult: data frame of global null tested results based on edges

Author(s)

Eric Minwei Liu, [e-mail]

Examples

data(netbox2010)

sifNetwork<-netbox2010$network
graphReduced <- networkSimplify(sifNetwork,directed = FALSE) 

geneList<-as.character(netbox2010$geneList)

results<-geneConnector(geneList=geneList,networkGraph=graphReduced,
                      pValueAdj='BH',pValueCutoff=0.05,
                      communityMethod='lec',keepIsolatedNodes=FALSE)

names(results)

# Suggested 100 iterations. 
# Use 5 interations in the exampel to save running time.
# globalTest <- globalNullModel(netboxGraph=results$netboxGraph, 
#                              networkGraph=graphReduced, 
#                              iterations=5, numOfGenes = 274)

mil2041/netboxr documentation built on May 20, 2023, 6:02 a.m.