#!/usr/bin/ruby

# Main program to compare ECIs of bulk with that of surface
#
# Written by k-fleak
#

require "#{ENV["HOME"]}/myRuby/parseSigmaSet.rb"
require "#{ENV["HOME"]}/myRuby/identifyCluster.rb"
require "#{ENV["HOME"]}/myRuby/parseECI.rb"

class ComparisonBulkSurfECI

  def initialize(sigmaSetBulk, sigmaSetSurf, coutBulk, coutSurf, eciBulk, eciSurf, multiplyCellBulk, multiplyCellSurf)
    
    sigmaSetBulkF = ParseSigmaSet.new(sigmaSetBulk)
    sigmaSetSurfF = ParseSigmaSet.new(sigmaSetSurf)
    #coutBulkF     = ParseClusterOutSort.new(coutBulk)
    #coutSurfF     = ParseClusterOutSort.new(coutSurf)
    eciBulkF      = ParseECI.new(eciBulk)
    eciSurfF      = ParseECI.new(eciSurf)
    parseCoutSurf = ParseClusterOutSearch.new(coutSurf)
    
    
    @numClusterBulk = sigmaSetBulkF.getNumCluster
    @numClusterSurf = sigmaSetSurfF.getNumCluster
    @eciBulkTmp     = eciBulkF.getECI
    @eciSurfTmp     = eciSurfF.getECI
    @indexRelation  = parseCoutSurf.getIndexRelation
    
    getECIperCluster(multiplyCellBulk, multiplyCellSurf)
    compareECI(coutBulk, coutSurf)
  end
  
  def printComparedECI
    @sameIndexBulkAndSurf.each do |sibs|
      indexBulk = sibs[0]
      indexSurfArray = sibs[1]
      print "#V(bulk) - "
      print indexBulk
      print "\n"
      indexSurfArray.each do |indexSurf|
        print @eciSurf[indexSurf]
        print "  "
        print "#V(surf) - "
        print indexSurf
        print "\n"
      end
      print @eciBulk[indexBulk]
      print "  "
      print "#V(bulk) - "
      print indexBulk
      print "\n"
      
      print "\n\n"
    end
  end
  
  
  private
  
  def getECIperCluster(multiplyCellBulk, multiplyCellSurf)
    @eciBulk = Hash.new()
    @eciBulkTmp.each do |etmp|
      numCluster = @numClusterBulk[etmp[0]]
      @eciBulk[etmp[0]] = etmp[1] / numCluster * multiplyCellBulk.to_f
    end
    @eciBulk.sort
    
    @eciSurf = Hash.new()
    @eciSurfTmp.each do |etmp|
      indexSortSurf = etmp[0]
      indexOrigSurf = @indexRelation[indexSortSurf]
      eciSortSurf   = etmp[1]
      numClusterAll = 0
      indexOrigSurf.each do |ios|
        numCluster = @numClusterSurf[ios.to_i] / multiplyCellSurf
        numClusterAll += numCluster
      end
      @eciSurf[indexSortSurf] = eciSortSurf / numClusterAll
    end
    @eciSurf.sort
  end
  
  def compareECI(coutBulk, coutSurf)
    @sameIndexBulkAndSurf = Hash.new
    @eciBulk.each do |eciB|
      indexBulk = eciB[0].to_s
      ids = IdentifyCluster.new(coutBulk, coutSurf, indexBulk)
      ids.start()
      indexSurf = Array.new
      unless ids.nBody == 0 then
        sameCluster = ids.getSameCluster
        case ids.nBody
        when 1
          sameCluster.sort.each do |sc|
            indexSurf.push(sc[1][1])
          end
        else
          sameCluster.each do |sc|
            indexSurf.push(sc[(sc.size)-1])
          end
        end
        @sameIndexBulkAndSurf[indexBulk.to_i] = indexSurf
      end
    end
  end
  
end


#comp = ComparisonBulkSurfECI.new("sigmaSet.out.bulk", "sigmaSet.out.surf", "cluster.out.sort.bulk", "cluster.out.sort.surf", "ECI.bulk.gene5.02", "ECI.gene5.02.surf", 1, 6)
comp = ComparisonBulkSurfECI.new(ARGV[0], ARGV[1], ARGV[2], ARGV[3], ARGV[4], ARGV[5], ARGV[6].to_f, ARGV[7].to_f)
comp.printComparedECI