#!/usr/bin/perl -w

use strict;

# get_ave_clone_band_size.pl, last update: 7/29/03
# gets the average clone size in bands
# by Chet Langin, clangin@siu.edu
# SIU Plant Biotechnology and Genome Core_facility

# get start time
use Time::HiRes;
my $start_time = Time::HiRes::time;

# check arguments
if(scalar(@ARGV) != 1) {
  print "Usage:  ./get_ave_clone_band_size.pl input_file\n";
  print "Example:  ./get_ave_clone_band_size.pl sorted_contig2clone_relations.txt\n";
  exit;
} # if
my $input_file_str = $ARGV[0];

use constant AVE_CLONE_LENGTH => 145000;

open(INPUT_FILE, "<", $input_file_str)
  or die "Cannot open input file:  $!\n";

my %lengths = ();
my $length_sum = 0;
my $length_divisor = 0;
while(<INPUT_FILE>) {
  chomp;
  my @fields = split /\t/;
  my $start = $fields[2];
  my $end = $fields[3];
  my $length = $end - $start;
  $length_sum += $length;
  $length_divisor++;
  if(exists($lengths{$length})) {
    $lengths{$length}++;
  } # if
  else {
    $lengths{$length} = 1;
  } # else
} # while
close INPUT_FILE;

# print the hash
foreach my $length (sort keys %lengths) {
  if($lengths{$length}) {
    print "$length:  $lengths{$length}\n";
  } # if
} # foreach

$length_divisor = 1 if ($length_divisor == 0);
my $ave_length = $length_sum / $length_divisor;
print "Average length:  $ave_length\n";
my $band_length = AVE_CLONE_LENGTH / $ave_length;
print "Average band length:  $band_length\n";

# get run time
my $run_time = Time::HiRes::time - $start_time;
print "Run time:  $run_time seconds\n";