#! /usr/bin/env Rscript
# -*- mode: R -*-
# Above tells emacs to go into R mode for editing (must be second line).

##
## chromCor:
##
##   Compute correlation of two jarched bed files along a chromosome (default chr19).
##
## To run this script, turn the execute bit on and run just as any other script file. The Rscript
## command must be in your path.  Rscript comes with R distributions 2.8 (?) and higher, and is in the
## same directory as the R script itself (eg, /usr/local/R/bin).
##
## Author: RET
## Date:  20 Aug 2009

usage <- paste('Usage:  chromCor [-c <chromosome>] [-n <column number>] [-z] [-u] <bed1.jarch> <bed2.jarch>',
               '',
               'Compute correlation of two bed files along a chromosome (default chr19).',
               '',
               'By default, <bed1.jarch> and <bed2.jarch> are jarched 5-column bed files, with score to be correlated in the',
               'fifth column.  There must be the same number of entries for the desired chromosome in each bed file.',
               '',
               '<chromosome> is a chromosome name; by default chr19, or "all", for all entries in the bed file.',
               '',
               '<column number> is the column in each file containing the data to correlate; by default, 5.',
               '',
               '-s specifies that bed files are starched instead of jarched.',
               '',
               '-z specifies that bed files are gzipped instead of jarched.',
               '',
               '-u specifies that bed files are uncompressed (not jarched or gzipped).',
               '',
               'Expects gchr and Rscript to be in the user\'s path.',
               '',
               'Writes results to stdout.',
               '',
               '\n',
               sep = '\n')

args <- commandArgs(T)

if(length(args) <= 1){
  cat(usage)
  quit('no')
}

chr <- 'chr19'
col <- 5
uncompr <- c('', '')

i <- 1
while(substr(args[i], 1, 1) == '-'){
  opt <- substr(args[i], 2, 2)
  if(opt == 'c'){
    chr <- args[i+1]
    if(chr == 'all') chr = ''
    i <- i + 2
  }else if(opt == 'n'){
    col <- as.integer(args[i+1])
    i <- i + 2
  }else if(opt == 'z'){
    uncompr <- c('zcat', 'zcat')
    i <- i + 1
  }else if(opt == 's'){
    uncompr <- c('unstarch', 'unstarch')
    i <- i + 1
  }else if(opt == 'u'){
    uncompr <- c('cat', 'cat')
    i <- i + 1
  }else{
    cat('Unrecognized option:', args[i], '\n', file = stderr())
    cat(usage, file = stderr())
    quit('no')
  }
}

bed <- args[c(i, i+1)]
rds <- list()

for(i in 1:2){
  if(uncompr[i] == ''){
    ext <- gsub('.*\\.', '', bed[i])
    if( ext == "bed"){
      uncompr[i] <- 'cat'
    } else if( ext == "jarch"){
      uncompr[i] <- 'gchr'
    } else if( ext == "starch"){
      uncompr[i] <- 'unstarch'
    } else if( ext == "gz" || ext == "zip"){
      uncompr[i] <- 'zcat'
    } else{
      cat('Unrecognized file extension:', bed[i], '\n', file = stderr())
      quit('no')
    }
  }

  if(chr == '')
    cmd <- paste(uncompr[i], '%BED')
  else{
    if(uncompr[i] == "cat" || uncompr[i] == "zcat"){
      cmd <- paste(uncompr[i], ' %BED | grep "^', chr, '" -', sep = '')
    }else if(uncompr[i] == "unstarch" || uncompr[i] == "gchr"){
      cmd <- paste(uncompr[i], chr, '%BED')
    }
  }

  cmd <- paste(cmd, ' | cut -f', col, sep = '')

  cat('Reading', bed[i], '...\n', file = stderr())

  con <- pipe(gsub('%BED', bed[i], cmd))
  rds[[i]] <- scan(con, what = numeric(0))
  close(con)
}

cv <- cor(rds[[1]], rds[[2]])
cat(cv, '\n')