##############################################################################################################
#  README for nPARS algorithm
# The nPARS function is an R-compliant function. 
# R is a free-downloadable software from http://www.r-project.org/
# 
#  Author: Yen-Yi Ho
#  Email: yho@umn.edu
#  Last Modified: Nov 10, 2011
#  Version: 1.0.1 
# License information: GNU GENERAL PUBLIC LICENSE version 3 @ http://www.gnu.org/licenses/gpl.html
# Dependency: DEAL 
####################################################################################################################


 

################ nPARS function usage############################################################################
Description: nPARS is a function for constructing five-node networks using network partition reassembly search (nPARS) algorithm. The nPARS algorithm consists of three main steps: (1) partition the network space to three-node (L,E,D) subnetworks, (2) selecting subnetworks and (3) assembling subnetwork into larger networks. The current version of the nPARS algorithm assemble two three-node networks into a five-node network. 

Usage: nPARS(Rdata=mydata, k1=10, k2=10)

Arguments: 
Rdata: Rdata is a list of data objects. The first list is the matrix for genotype data with each row as a different individual and each column as a separate snp locus. The second list is the expression matrix with each row as a different individual and each column as a seperate expression measure. The thrid list contain a vector of one phenotype outcome.
k1: Select the top k1 three-node subnetworks with largest phi scores for reassembly. Default is 10.  
k2: Select the top k2 five-node subnetworks with largest phi score to report. Default is 10. 

Value: 
	nPARS returns a list of objects of class network. Methods for printing and plotting are defined. 

Reference: Yen-Yi Ho, Leslie M. Cope, Giovanni Parmigiani. A Comprehensive Search Algorithm for Constructing Bayesian Networks Using Large-Scale Genomic Data. (Submitted, Available upon request)

See also: network-class (deal R package)

Examples: 

######################
Example 1: 

library(deal)
source("nPARSsource2.R")
PubL<-read.delim("PubL.txt", header=TRUE, sep="\t")
PubE<-read.delim2("PubE.txt", header=TRUE, sep="\t")
PubD<-read.delim("PubD.txt", header=TRUE, sep="\t")
mydata<-NULL
mydata[[1]]<-PubL[,6:15]
mydata[[2]]<-PubE[,6:15]
mydata[[3]]<-as.numeric(PubD[,7])
mydata[[1]]<-apply(mydata[[1]], 2, as.numeric)
mydata[[2]]<-apply(mydata[[2]],2,as.numeric)
mydata[[3]]<-as.numeric(mydata[[3]])
Rdata<-mydata
## Running nPARS algorithm
finalNet<-nPARS(Rdata=mydata, k1=10, k2=10)
print(finalNet)
plot(finalNet[[1]])

############################# End of Example