#This scrpit describe how to build a global network from KEGG database files using the KEML2igraph function from package NetPathMiner
library(igraph)
library(NetPathMiner)
library(org.Hs.eg.db)

#load all the kegg human pathways into one igraph object
path.db<-"Path to KEGG files" #modify this path according to your own path
files<-list.files(path.db)
setwd(path.db)
graph<-KGML2igraph(filename=files,parse.as = "signaling",expand.complexes = TRUE)
#convert id to gene symbol
g.ft<-formatGraph(graph)
#remove the vertices which could not be converted to entrez gene id 
#remove the vertices which have no links to others
g.fm<-filterGraph(g.ft)
#remove the loops
g.dl<- simplify(g.fm, remove.multiple = F, remove.loops = T)
save(g.dl,file="keggAllPath.RData")

#Functions needed
formatGraph<-function(tmp.graph){ #convert id to symbol
  v<-V(tmp.graph)
  v.entrez<-as.vector(names(v))
  v.entrez<-gsub("hsa:","",v.entrez)
  v.sym<-mget(v.entrez,org.Hs.egSYMBOL,ifnotfound = NA)
  tmp.num<-sum(ifelse(is.na(v.sym),1,0))
  if(tmp.num==0){
    print("All entrez ids were converted to symbols.")
  }else{
    print(paste(tmp.num," entrez ids were not converted to symbols.",sep=""))
    v.sym[is.na(v.sym)]<-"NotGene"
  }
  graph.sym <- set.vertex.attribute(tmp.graph, "label", value=v.entrez)
  graph.sym <- set.vertex.attribute(tmp.graph, "name", value=v.sym)
  graph.sim<-simplify(graph.sym, remove.multiple = T, remove.loops = T)
  return(graph.sim)
}

filterGraph<-function(graph,deg=0,unmapped="NotGene"){ #remove vertices which failed in id conversion, and remove isolated genes
  tmp.deg <- degree(graph, mode="all")
  n<-length(tmp.deg[names(tmp.deg) == unmapped])
  graph.rm<-graph
  if(n>0){
    for(i in 1:n){
      graph.rm<-delete.vertices(graph.rm,"NotGene")
    }
  }
  deg.in<-colSums(as.matrix(graph.rm[]))
  deg.out<-rowSums(as.matrix(graph.rm[]))
  deg.new<-deg.in+deg.out
  v.rm<-names(deg.new[deg.new==deg])
  graph.rm.2<-delete_vertices(graph.rm,v.rm)
  return(graph.rm.2)
}