# Begin script
#-------------
#!/usr/bin/ruby

require 'rubygems'
require 'ruby-debug' #For debugging only

#7/17/12  Evalue Picker
#This program will take an output from blaster and bin the E-values into orders of magnitude and then out put the number of hits
#that occur with an evalue hit of that order of magnitude.

blaster = File.open("GenDB_Blaster_Output.txt")

#define variables for later use
hitlist = []


=begin
  For the Eval list, I chose to store them in a key value pair (array)


  So it would look like

   array = {
    "6e10" => "1"
    "4e10" => "2"
   }

  That way I could check if the key existed, if not, create the entry, otherwise just add one to the keys value
=end

Eval_list = {}

#Each line of blaster file create the variable named bar
blaster.each { |bar|
  bar.chomp! #Strip white space

  fields_from_blaster = bar.split("\t", 12)

  seqh = {
    'contig'    => fields_from_blaster[0],
    'hit'       => fields_from_blaster[1],
    '%ID'       => fields_from_blaster[2],
    'len'       => fields_from_blaster[3],
    '#mismatch' => fields_from_blaster[4],
    '#gaps'     => fields_from_blaster[5],
    'qstart'    => fields_from_blaster[6],
    'qend'      => fields_from_blaster[7],
    'hstart'    => fields_from_blaster[8],
    'hend'      => fields_from_blaster[9],
    'Eval'      => fields_from_blaster[10],
    'score'     => fields_from_blaster[11]
  }

  hitlist << seqh

  #determine if an evalue has already been in the array

  evalue = seqh["Eval"] #shortcut

  if Eval_list[evalue].nil? #If there is NOT an entry for this evalue, we want to add it
    Eval_list[evalue] = 1
  else #There is an entry for this evalue so we need to add one to its value
    Eval_list[evalue] = Eval_list[evalue] + 1
  end
}

Eval_list.each do |ev, freq|
puts "#{ev}\t#{freq}\n"
end