getNetwork: Building Networks of Nodes

In BioTIP: BioTIP: An R package for characterization of Biological Tipping-Point

Description

This function builds one correlation network for each state (sample group) and runs across all states. The network nodes are defined by the context of the input dataset. For transcriptomic network analysis, network nodes can be the expressed transcript IDs and network links can be the correlation coefficients. Using the Pearson Correlation Coefficient (PCC) analysis, this function assembles a correlation network of nodes (e.g., co-expressed transcripts) for each state using the R package igraph.

Usage

getNetwork(optimal, fdr = 0.05)

Arguments

optimal	A list of x numeric data frames, where x is the number of states studied. Each data frame consists of loci with high standard deviations. This object can be obtained through sd_selection function.
fdr	A numeric cutoff value for a Pearson Correlation Coefficient (PCC) analysis. Default is 0.05. Transcripts are linked into a network if their correlations meet this PCC-significance criterion.

Value

A list of igraph objects whose length is the length of the input object optimal. Each object is a network of correlated nodes whose PCCs meet the significant criteria based on the false discovery rate (FDR) control. The length of the list is the number of states with PCC networks. If no PCC meets the significant criteria in a state, the state will be deleted from the output.

Author(s)

Zhezhen Wang [e-mail]; Xinan H Yang [e-mail]

Examples

test = list('state1' = matrix(sample(1:10, 6), 2, 3), 
 'state2'=matrix(sample(1:10, 6), 2, 3), 
 'state3' = matrix(sample(1:10, 6), 2, 3))

for(i in names(test)){
  colnames(test[[i]]) = 1:3
  row.names(test[[i]]) = 1:2}

igraphL <- getNetwork(test,  fdr=1)
#[1] "state1:2 nodes"
#[1] "state2:2 nodes"
#[1] "state3:2 nodes

Example output

state1:2 nodes
state2:2 nodes
state3:2 nodes

BioTIP documentation built on Nov. 8, 2020, 6:27 p.m.