ave.chip.r

# feature.unsorted <- read.table(file="feature.bed")
# data.unsorted <- read.table(file="data.log.enrichment.sgr")
# starts with log2 enrichment values, converts to enrichment and normalizes to number of data points in each feature
# sort feature and sgr by chromosome arm:
feature <- feature.unsorted[order(feature.unsorted$V1), ]
data <- data.unsorted[order(data.unsorted$V1), ]
length.feature<-dim(feature)[1]
feature.raw <- rep(NA, length.feature)
feature.value <- rep(NA, length.feature)
feature.arm<-feature[ ,1]
feature.start<-feature[ ,2]
feature.end <- feature[ ,3]
length.data<-dim(data)[1]
data.arm <- data[ ,1]
data.position <- data[ ,2]
data.value <- 2^data[ ,3]
i <- 1
j <- 1
k <- 1
x <- 0
for(i in 1:length.feature){
  sum <- 0
  count <- 0
  start <- feature.start[i]
  end <- feature.end[i]
  for(j in k:length.data){
    if(data.arm[j]==feature.arm[i]){
      if(data.position[j]<start){next}
      if(data.position[j]>end){break}
      sum <- sum+data.value[j]
      count <- count+1
      x <- j
    }
    else{next}
  }
  k <- x-count
  if(k < 1){k <- 1}
  feature.raw[i] <- sum
  if(feature.raw[i]==0){feature.raw[i] <- 1}
  feature.value[i] <- sum / count  
}
log.feature.value <- log2(feature.value)
feature.data <- cbind(feature, feature.value, log.feature.value)
write.table(feature.data, file="feature_data.txt", quote=FALSE, sep="\t", row.names=FALSE, col.names=FALSE)
# end

sum.reads.r

# feature.unsorted <- read.table(file="feature.bed")
# data.unsorted <- read.table(file="data.sgr") - is sgr file giving number of read at genomic positions
# sums data across feature, adds offset of 1 to all features to preven 0 values
# sort feature and sgr by chromosome arm:
feature <- feature.unsorted[order(feature.unsorted$V1), ]
data <- data.unsorted[order(data.unsorted$V1), ]
length.feature<-dim(feature)[1]
feature.value<-rep(NA,length.feature)
feature.arm<-feature[ ,1]
feature.start<-feature[ ,2]
feature.end <- feature[ ,3]
length.data<-dim(data)[1]
data.arm <- data[ ,1]
data.position <- data[ ,2]
data.value <- data[ ,3]
i <- 1
j <- 1
k <- 1
x <- 0
for(i in 1:length.feature){
  sum <- 0
  count <- 0
  start <- feature.start[i]
  end <- feature.end[i]
  for(j in k:length.data){
    if(data.arm[j]==feature.arm[i]){
      if(data.position[j]<start){next}
      if(data.position[j]>end){break}
      sum <- sum+data.value[j]
      count <- count+1
      x <- j
    }
    else{next}
  }
  k <- x-count
  if(k < 1){k <- 1}
  feature.value[i] <- sum + 1
}
log.feature.value <- log2(feature.value)
feature.data <- cbind(feature, feature.value, log.feature.value)
write.table(feature.data, file="feature_data.txt", quote=FALSE, sep="\t", row.names=FALSE, col.names=FALSE)
# end