#!/usr/bin/perl -w

use strict;

# extract_loci.plx, last update: 9/2/03
# extract locus anchor data from a file provided by Iowa State
# by Chet Langin, clangin@siu.edu
# SIU Plant Biotechnology and Genome Core-facility

# begin the timer
use Time::HiRes;
my $start_time = Time::HiRes::time;

# check the arguments
if(scalar(@ARGV) > 0) {
  print "The arguments are preset as follows:\n";
  print "  Input file:  raw_loci.txt\n";
  print "  Output file:  locus_anchors.txt\n";
  print "  Sorted output file:  sorted_locus_anchors.txt\n";
  print "Restart without any arguments to continue...\n";
  exit;
} # if

use constant FACTOR => 441824.931130786;

# set the filenames
my $input_file_str = "raw_loci.txt";
my $output_file_str = "locus_anchors.txt";
my $sorted_output_file_str = "sorted_locus_anchors.txt";

# read the input file
open(INPUT_FILE, "<", $input_file_str)
  or die "Cannot open input file:  $!\n";
my @file_array = <INPUT_FILE>;
close INPUT_FILE;
my $number_lines = scalar(@file_array);

my $stable_label_name = "";
my $current_label_name = "";
my $current_mlg = "";
my $current_cm = -1;
my $current_midpoint;
open(OUTPUT_FILE, ">", $output_file_str)
  or die "Cannot open output file:  $!\n";
# look at each line in the input file
for(my $current_line=0; $current_line < $number_lines; $current_line++) {
  if($file_array[$current_line] =~ / : /) {
    my @field_array = split /"/, $file_array[$current_line];
    $current_label_name = $field_array[1];
    if($field_array[0] =~ /Locus/) {
      if($current_mlg eq "" || $current_cm == -1) {
        print "$current_line:  MLG or cM problem\n";
      } # if
      else {
        $current_midpoint = 101 if($current_midpoint < 101);
        my $output_str = sprintf("%s\t%08d\t%08d\t%08d\t%s\n", $current_mlg, $current_midpoint - 100, $current_midpoint + 100, $current_midpoint, $stable_label_name);
        print OUTPUT_FILE $output_str;
      } # else
      $stable_label_name = $current_label_name;
      $current_mlg = "";
      $current_cm = -1;
    } # if
  } # if
  elsif($file_array[$current_line] =~ /^Map.+Composite_2/) {
    # clean up the end of the line
    chomp $file_array[$current_line];
    chop $file_array[$current_line];
    $file_array[$current_line] =~ s/\s*$//;

    if($current_mlg ne "") {
      print "More than one location:  $current_line\n";
    } # if

    # get the MLG
    $file_array[$current_line] =~ /"(.+)-/;
    $current_mlg = $1;
    $current_mlg = "D1AQ" if($current_mlg eq "D1a");
    $current_mlg = "D1BW" if($current_mlg eq "D1b");

    # get the cM location, change it to bases (round it)
    my @field_array = split / /, $file_array[$current_line];
    $current_cm = $field_array[scalar(@field_array) - 1];
    $current_midpoint = sprintf("%.0f", $current_cm * FACTOR);
  } # elsif
} # for
# close OUTPUT_FILE;

system("sort $output_file_str > $sorted_output_file_str");

# my $run_time = Time::HiRes::time - $start_time;
print "Run time:  ${\(Time::HiRes::time - $start_time)} seconds\n";