# efg.  Stowers Institute.
# 7 Feb 2006.  Updated 14 May 2006.

sink("2-ExtractExpressionData.txt")

# Patient Order (Load all patients)
filename <- "C:/CAMDA06/1-Clinical/Classification.csv"
patient <- read.csv(filename, as.is=TRUE)
dim(patient)                                     # 227 8
patient$order <- 1:nrow(patient)

# Gene/Probe Order (List of only probes of interest)
basepath <- "C:/CAMDA06/5-GeneExpressionAnalysis/b-PsychoNeuroendocrineImmune/"
filename <- paste(basepath, "ProbeListUnique.csv", sep="")
probe <- read.csv(filename, as.is=TRUE)
dim(probe)                                       # 1917 10
probe$order <- 1:nrow(probe)

# Expression Data [Five "extra" arrays have already been excluded]
basepath <- "C:/CAMDA06/4-GeneExpressionScaling/"
filename <- paste(basepath, "PatientScaledExpression.csv", sep="")
Expression <- read.csv(filename, as.is=TRUE)
colnames(Expression)[1] <- "ID"
dim(Expression)                                  # 172 19701

Expression.match.ID <- as.numeric(substr(Expression$ID,1,8))
length(Expression.match.ID)                      # 172

PatientMatch <- patient$ABTID %in% Expression.match.ID
patient <- patient[PatientMatch,]
patient$order <- 1:nrow(patient)
write.csv(patient, file="ArrayPatients.csv", row.names=FALSE)
dim(patient)                                     # 164 9  (8 duplicates)

# For all Gene Expression patients, establish CLUSTER order
for (i in 1:nrow(Expression))
{
  matchid <- as.numeric(substr(Expression$ID[i],1,8))
  if (length( which(matchid == patient$ABTID) ) == 0)
  {
    Expression$Order[i] <- 0  # Should never happen
  } else {
    Expression$Order[i] <- patient$order[ which(matchid == patient$ABTID) ]
  }
}

Expression <- Expression[order(Expression$Order, Expression$ID),]
dim(Expression)                                  # 172 19702

# Find location of the 8 replicates.  Label these "A" and +1 then "B"
DuplicateOrder <-  which(Expression$Order[1:(length(Expression$Order)-1)] ==
                         Expression$Order[2:length(Expression$Order)])

# List original filenames so "A" and "B" can be resolved later if necessary
Expression$ID[DuplicateOrder]
Expression$ID[DuplicateOrder+1]

ABTID  <- substr(Expression$ID,1,8)

# Verify duplicates
ABTID[DuplicateOrder]
ABTID[DuplicateOrder+1]

ABTID[DuplicateOrder]   <- paste(ABTID[DuplicateOrder],   "A", sep="")
ABTID[DuplicateOrder+1] <- paste(ABTID[DuplicateOrder+1], "B", sep="")

# Verify duplicates
ABTID[DuplicateOrder]
ABTID[DuplicateOrder+1]

rownames(Expression) <- ABTID

# get rid of these columns

Expression$Order <- NULL
Expression$ID   <- NULL

dim(Expression)                                  # 172 19700

# Order probes by category/subcategory
ProbeOrder  <- rep(0,  ncol(Expression))
ProbeGene   <- rep("", ncol(Expression))
ProbeName   <- rep("", ncol(Expression))
ProbeSuffix <- rep("", ncol(Expression))
ProbeSystem  <- rep("", ncol(Expression))

# Establish gene / probe order based on PNI list
for (j in 1:ncol(Expression))
{
  matchid <- colnames(Expression)[j]

  # In R the probes like "X (1)" became "X..1."
  parts <- strsplit(matchid, "\\.\\.")
  matchid <- parts[[1]][1]

  if (length(parts[[1]]) == 1)
  {
    suffix <- ""
  } else {
    suffix <- paste(".", substr( parts[[1]][2],1,1), sep="")
  }

  index <- which(matchid == probe$probe)
  if (length(index) > 0)
  {
    ProbeOrder[j]  <- probe$order[index]
    ProbeGene[j]   <- probe$gene[index]
    ProbeName[j]   <- matchid
    ProbeSuffix[j] <- suffix
    ProbeSystem[j]  <- probe$System[index]
  }
}

# Get rid of blank columns
NonZeroColumns <- ProbeOrder > 0
Expression <- Expression[, NonZeroColumns]
ProbeGene <- ProbeGene[NonZeroColumns]
ProbeName <- ProbeName[NonZeroColumns]
ProbeSuffix <- ProbeSuffix[NonZeroColumns]
ProbeSystem <- ProbeSystem[NonZeroColumns]
ProbeOrder <- ProbeOrder[NonZeroColumns]

# Reorder everything now by order(ProbeOrder)
Expression <- Expression[,order(ProbeOrder)]
ProbeGene <- ProbeGene[order(ProbeOrder)]
ProbeName <- ProbeName[order(ProbeOrder)]
ProbeSuffix <- ProbeSuffix[order(ProbeOrder)]
ProbeSystem <- ProbeSystem[order(ProbeOrder)]
ProbeOrder <- ProbeOrder[order(ProbeOrder)]

colnames(Expression) <-  paste(ProbeGene, ".", ProbeName, ProbeSuffix, sep="")
dim(Expression)     # 172 1914

write.csv(Expression, file="SelectedProbes.csv")

ProbeListAll <- data.frame(ID=colnames(Expression),
                           Gene=ProbeGene,
                           Name=ProbeName,
                           Suffix=ProbeSuffix,
                           System=ProbeSystem,
                           Unique=ProbeOrder)
ProbeListAll <- cbind(ProbeListAll, probe[ProbeListAll$Unique,])
ProbeListAll <- ProbeListAll[, -c(6,7,10,17)]
write.csv(ProbeListAll, file="ProbeListAll.csv")

sink()