# 12 Feb 2011, 12h00

#library(ALL)
library(ConsensusClusterPlus)

##############################################################################
# cd /projects/remc_bigdata/TCGA/LAML/consensus_clustering/20110417
# source("/projects/remc_bigdata/TCGA/LAML/consensus_clustering/20110417/ConsensusClusterPlus.LAML-isomiRs.xhost09.20110417.R")

output.folder <- "/projects/remc_bigdata/TCGA/LAML/consensus_clustering/20110417"

# data
cat("\nread data\n")

# 143 isomir-maturestar (17 April, 09h13)
#mydata <- read.table(file="/projects/remc_bigdata/TCGA/LAML/expn_matrix_norm_passed_TCGA.180-LAML.143-isomir-maturestar-q80.20110416.txt", header = TRUE, row.names = 1, sep="\t")
#subfolder.title <- "AML-180.143-isomir-maturestar-q80.km_spearman_reps500"

# 98 mir-maturestar (17 April, 09h12a)
#mydata <- read.table(file="/projects/remc_bigdata/TCGA/LAML/expn_matrix_norm_passed_TCGA.180-LAML.98-mir-maturestar-q80.20110416.txt", header = TRUE, row.names = 1, sep="\t")
#subfolder.title <- "AML-180.98-mir-maturestar-q80.km_spearman_reps500"

# 111 mature-strands (17 April, 09h11b)
#mydata <- read.table(file="/projects/remc_bigdata/TCGA/LAML/expn_matrix_norm_passed_TCGA.180-LAML.111-mature-strands-q80.20110416.txt", header = TRUE, row.names = 1, sep="\t")
#subfolder.title <- "AML-180.111-mature-strands-q80.km_spearman_reps500"

# 163 abs>100, frac.re>0p1, isomiRs (17 April, 09h11a)
#mydata <- read.table(file="/projects/remc_bigdata/TCGA/LAML/expn_matrix_norm_passed_TCGA.180-LAML.163-a100-fre0p1-isomiRs-q80.20110416.txt", header = TRUE, row.names = 1, sep="\t")
#subfolder.title <- "AML-180.163-a100-fre0p1-isomiRs-q80.km_spearman_reps1000"

# 329 isomirs  (17 April, 09h15)
#mydata <- read.table(file="/projects/remc_bigdata/TCGA/LAML/expn_matrix_norm_passed_TCGA.180-LAML.329-expr-isomiRs-q80.20110416.txt", header = TRUE, row.names = 1, sep="\t")
#subfolder.title <- "AML-180.329-expr-isomiRs-q80.km_spearman_reps500"



dim(mydata)
mydata.matrix <- as.matrix(mydata)
cat("\nset directories\n")
setwd(output.folder)
output.subfolder <- paste( output.folder,"/", subfolder.title,"/", sep="" )
icl.string <- "LAML-180.20110416.txt"

##############################################################################
kmax <- 12

cat("\nrun ConsensusClusterPlus()\n")
results = ConsensusClusterPlus( mydata.matrix, maxK=kmax, reps=1000, pItem=0.8, pFeature=1, title=subfolder.title, clusterAlg="km", distance="spearman", seed=1262118388.71279, plot="pdf", writeTable=T, verbose=T )


# Identify an interesting clustering result
#cT <- results[[17]][["consensusTree"]]
# tree order
#cTorder <- cT$order
#write.table( cTorder, file="cT17order.LAML.DNAmeth.hclust_beta_spearman_q50_noNA_reps100.BRCA.20110113.txt",sep="\t")

# a function to take an integer k value and write out an order file (13 Jan 2010)
writeOrder <- function( k, results ){
	cT <-       results[[k]][["consensusTree"]]
	cTorder <-  cT$order
	pathfile <- paste( output.subfolder, "cTorder.k", k, ".txt", sep="" )
	write.table( cTorder, file=pathfile, sep="\t")
}
cat("\nwriting 'item order' files for each clustering result (K value)\n")
for(k in 2:kmax) writeOrder(k,results)

#results[[2]][["consensusTree"]]



##############################################################################
cat("\nrun calcICL()\n")
icl = calcICL( results, title=subfolder.title, plot="pdf", writeTable=T )
icl.cC <- icl[["clusterConsensus"]]
icl.iC <- icl[["itemConsensus"]]

cat("\nwriting calcICL() files\n")
setwd(output.subfolder)
write.table( icl, file=paste( output.subfolder, "icl.", icl.string, sep="" ), sep="\t" )
write.table( icl.cC, file=paste( output.subfolder, "icl.clusterConsensus.", icl.string, sep="" ), sep="\t" )
write.table( icl.iC, file=paste( output.subfolder, "icl.itemConsensus.", icl.string, sep="" ), sep="\t" )