R/fetchPtCorrGrinnNetwork.R

In kwanjeeraw/grinn: graph database and R package for omics data integration

#'Compute a partial correlation network and expand the network with information from Grinn internal database
#'@description from input omics data e.g. normalized expression data or metabolomics data, it is a one step function to:
#'
#'1. Compute a partial correlation network of input omics data using qpgraph functions. 
#'The correlation coefficients are continuous values between -1 (negative correlation) and 1 (positive correlation), with numbers close to 1 or -1, meaning very closely correlated.
#'
#'2. Expand the correlation network using information from the Grinn internal database.
#'The nodes of the correlation network are the keywords input to query the Grinn internal database.
#'The Grinn internal database contains the networks of the following types that can get expanded to: 
#'metabolite-protein, metabolite-protein-gene, metabolite-pathway, protein-gene, protein-pathway and gene-pathway, see also \code{\link{fetchGrinnNetwork}}.
#'@usage fetchPtCorrGrinnNetwork(datX, corrCoef, pval, alpha, epsilon, matrix.completion, returnAs, xTo, filterSource)
#'@param datX data frame containing normalized, quantified omics data e.g. expression data, metabolite intensities. 
#'Columns correspond to entities e.g. genes, metabolites, and rows to samples e.g. normals, tumors. 
#'Require 'nodetype' at the first row to indicate the type of entities in each column. See below for details.
#'@param corrCoef numerical value to define the minimum value of absolute correlation, from 0 to 1, to include edges in the output.
#'@param pval numerical value to define the maximum value of pvalues, to include edges in the output.
#'@param alpha a numeric value specifying significance level of each test used in \code{\link{qpAvgNrr}}.
#'@param epsilon a numeric value specifying the maximum cutoff value of the non-rejection rate met by the edges that are included in the qp-graph, see \code{\link{qpGraph}}.
#'@param matrix.completion a string specifying algorithm to employ in the matrix completion operations used in \code{\link{qpPAC}}
#'@param returnAs string of output type. Specify the type of the returned network. 
#'It can be one of "tab","json","cytoscape", default is "tab". "cytoscape" is the format used in Cytoscape.js
#'@param xTo string of node type. It can be one of "metabolite","protein","gene","pathway". See below for details.
#'@param filterSource string or list of pathway databases. The argument is required, if \code{xTo = "pathway"}.
#'The argument value can be any of "SMPDB","KEGG","REACTOME" or combination of them e.g. list("KEGG","REACTOME").
#'@details datX is matrix in which rows are samples and columns are entities. 
#'
#'- The correlation network can be expand from datX entites to a specific nodetype, by providing a value to \code{xTo}.
#'
#'If \code{xTo} is given, the columns of datX are required to use grinn ids for extended queries on the Grinn internal database, see \code{\link{convertToGrinnID}} for id conversion.
#'
#'If \code{xTo} = NULL , only the correlation network will be returned.
#'@return list of nodes and edges. The list is with the following componens: edges and nodes. Return empty list if found nothing
#'@author Kwanjeera W \email{kwanich@@ucdavis.edu}
#'@references Castelo R. and Roverato A. (2006) A robust procedure for Gaussian graphical model search from microarray data with p larger than n. J. Mach. Learn. Res., 7:2621-2650.
#'@references Castelo R. and Roverato A. (2009) Reverse engineering molecular regulatory networks from microarray data with qp-graphs. J Comput Biol, 16(2), pp. 213-27.
#'@export
#'@seealso \pkg{\link{qpgraph}}, \code{\link{qpAvgNrr}}, \code{\link{qpGraph}}, \code{\link{qpPAC}}, \code{\link{fetchGrinnNetwork}}, \url{http://js.cytoscape.org/}
#'@examples
#'# Compute a partial correlation network of metabolites and expand to a grinn network of metabolite-protein
#'dummy <- rbind(nodetype=rep("metabolite"),t(mtcars))
#'colnames(dummy) <- c('G1.1','G27967','G371','G4.1',paste0('G',sample(400:22000, 28)))
#'result <- fetchPtCorrGrinnNetwork(datX=dummy, corrCoef=0.7, pval=0.05, returnAs="tab", xTo="protein")
#'library(igraph)
#'plot(graph.data.frame(result$edges[,1:2], directed=FALSE))

fetchPtCorrGrinnNetwork <- function(datX, corrCoef=0.5, pval=0.05, alpha=0.05, epsilon=0.5, matrix.completion="IPF", returnAs="tab", 
                                  xTo=NULL, filterSource=list()){
  corrnw = fetchPtCorrNetwork(datX=datX,corrCoef=corrCoef,pval=pval,alpha=alpha,epsilon=epsilon,matrix.completion=matrix.completion,returnAs="tab")
  if(nrow(corrnw$nodes)>0){
    nodetypes = tolower(unique(corrnw$nodes$nodetype))
    if(length(nodetypes)>1){#if there are two data types
      if(!is.null(xTo)){
        basicnw4 = fetchGrinnNetwork(txtInput=corrnw$nodes[which(tolower(corrnw$nodes$nodetype)==nodetypes[1]), 1],from=nodetypes[1],
                                     to=xTo,filterSource=filterSource,dbXref="grinn") #relations sources to others
      }else{
        basicnw4 = list(nodes=data.frame(),edges=data.frame())
      }
      basicnwNodes = basicnw4$nodes #collect all nodes
      basicnwEdges = basicnw4$edges #collect all edges
      if(nrow(basicnwEdges)>0){
        basicnwEdges = basicnwEdges[!duplicated(basicnwEdges[,1:ncol(basicnwEdges)]),] #remove duplicated edges
        basicnwNodes = basicnwNodes[!duplicated(basicnwNodes[,1]),] #remove duplicated nodes
        basicnw = list(nodes=basicnwNodes, edges=basicnwEdges)
      }else{
        basicnw = list(nodes=data.frame(),edges=data.frame())
      }
    }else{#if there is only one data type
      if(!is.null(xTo)){
        basicnw = fetchGrinnNetwork(txtInput=corrnw$nodes$id,from=nodetypes,to=xTo,filterSource=filterSource,dbXref="grinn")
      }else{
        basicnw = list(nodes=data.frame(),edges=data.frame())
      }
    }
    #collect node info
    corrattb = data.frame()
    corrattb = plyr::ldply (apply(corrnw$nodes, MARGIN = 1, FUN=getNodeInfo, x = "id", y = "nodetype")) #format nodelist
    corrnw$edges$source = lapply(corrnw$edges$source, FUN=formatId, y = corrattb) #format edgelist
    corrnw$edges$target = lapply(corrnw$edges$target, FUN=formatId, y = corrattb) #format edgelist
    if(nrow(basicnw$nodes)>0){
      cat("Formating and returning combined network ...\n")
      basicnw$edges$corr_coef = 1
      basicnw$edges$pval = 0
      basicnw$edges$direction = 0
      corrnw$edges$relsource = ""
      corrnw$nodes$xref = ""
      corrnw$nodes$gid = corrnw$nodes$id #same ids
      pair = rbind(basicnw$edges,corrnw$edges)
      if(nrow(corrattb)>0){attb = rbind(basicnw$nodes,corrattb,corrnw$nodes)}else{attb = rbind(basicnw$nodes,corrnw$nodes)}
      attb = attb[!duplicated(attb[,2]),]
      cat("Found ",nrow(pair)," relationships...\n")
    }else{#if only correlation network found
      cat("Formating and returning combined network ...\n")
      pair = corrnw$edges
      corrnw$nodes$xref = ""
      corrnw$nodes$gid = corrnw$nodes$id #same ids
      if(nrow(corrattb)>0){attb = rbind(corrattb,corrnw$nodes)}else{attb = corrnw$nodes}
      attb = attb[!duplicated(attb[,2]),]
      cat("Found ",nrow(pair)," relationships...\n")
    }
  }else{#if no correlation network found
    print("Returning no data...")
    pair = data.frame()
    attb = data.frame()
    cynetwork = list(nodes="", edges="")
  }
  out = switch(returnAs,
               tab = list(nodes=attb, edges=pair),
               json = list(nodes=jsonlite::toJSON(attb), edges=jsonlite::toJSON(pair)),
               cytoscape = createCyNetwork(attb, pair),
               stop("incorrect return type"))
}