#!/usr/bin/perl -w

use strict;

# extract_loci2.plx, last update: 9/14/03
# extract locus anchor data from a USDA spreadsheet
# 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:  usda.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 = "usda.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 $name = "";
my $mlg = "";
my $cm = -1;
my $midpoint = 0;
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++) {
  my @field_array = split /\t/, $file_array[$current_line];
  $name = $field_array[0];
  $mlg = $field_array[1];
  $mlg = "D1AQ" if($mlg eq "D1a");
  $mlg = "D1BW" if($mlg eq "D1b");
  $cm = $field_array[2];
  $midpoint = sprintf("%.0f", $cm * FACTOR);
  $midpoint = 101 if($midpoint < 101);
  my $output_str = sprintf("%s\t%08d\t%08d\t%08d\t%s\n", $mlg, $midpoint - 100, $midpoint + 100, $midpoint, $name);
  print OUTPUT_FILE $output_str;
} # 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";