GA_search_connected: Genetic algorithm for module identification on weighted...

In fairmiracle/PPINet: Protenin-protein interaction network construction and module identification

Description

Using genetic algorithm (GA) to identify active modules on weighted network. The connectedness of resulted module is ensured by connected components finding of binary encoded set of genes. Rewritten from COSINE package.

Usage

GA_search_connected(lambda, node_score, edge_score, EdgeList, num_iter = 1000,
  muCh = 0.05, zToR = 10, minsize = 30)

Arguments

lambda	weight parameter in the objective
node_score	A N-length vector storing the nodes
edge_score	A M-length vector storing the edges
EdgeList	A two-columns matrix with each row a pair of weighted edges
num_iter	Number of iterations in GA
muCh	Mutation rate in GA
zToR	Zero to one ratio
minsize	The minimal size of module

Value

a list containing module size, best score, module as a list of nodes and a list of GA functions.

Author(s)

Dong Li, [e-mail]

References

Ma, Haisu, et al. "COSINE: COndition-SpecIfic sub-NEtwork identification using a global optimization method." Bioinformatics 27.9 (2011): 1290-1298.

Dong Li et al. Memetic algorithm for finding active connected subnetworks in intracellular networks. 2016.

Examples

library(COSINE)
data(scaled_node_score)
data(scaled_edge_score)
data(PPI)
GA_result <- GA_search_connected(lambda=0.5,scaled_node_score,
scaled_edge_score,PPI,num_iter=100, muCh=0.05, zToR=10, minsize=50)
GA_result2<-GA_search_PPI(lambda=0.5,scaled_node_score,scaled_edge_score,PPI, 
num_iter=100, muCh=0.05, zToR=10, minsize=50)
## visualized by igraph
selected = c()
for (i in 1:dim(PPI)[1]){
    if (PPI[i,1] %in% GA_result$Subnet || PPI[i,2] %in% GA_result$Subnet)
        selected = c(selected,i)
}
library(igraph)
g <- graph.data.frame(PPI[selected,], directed=FALSE)
V(g)$color <- "gray"
V(g)$color[match(GA_result$Subnet,V(g)$name)] <- "red"
layout <- layout.reingold.tilford(g, circular=T)
png(filename = 'Connected_PPI.png', width=4096, height = 3652)
plot(g,layout=layout, vertex.size=5, vertex.label.cex=5,vertex.label.dist=0.5,edge.width=5)
dev.off()

fairmiracle/PPINet documentation built on May 16, 2019, 9:59 a.m.