#!/usr/bin/perl -w

use strict;

# extract_annotation.plx, last update:  11/7/03
# extracts annotations from a file that Jeff provides and puts it into
#   a relation file which I can use.
# by Chet Langin, clangin@siu.edu
# SIU Plant Biotechnology and Genomics Core-facility

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

my $input_str = "annotation.txt";
my $output_str = "clone2comment.txt";

my %comment_hash = ();

open(INPUT, "<", $input_str)
  or die "Cannot open input file:  $!\n";
my @file_array = <INPUT>;
close INPUT;

# making a hash of all of the comments for a clone
my $file_length = scalar(@file_array);
for(my $lcv = 0; $lcv < scalar(@file_array); $lcv++) {
  my $current_line = $file_array[$lcv];
  next if($current_line =~ /reference/);
  chomp $current_line;
  my @line_array = split / /, $current_line;
  my $relation = $line_array[1];
  my @feature_array = split /_/, $relation;
  my $clone_name = $feature_array[0];
  next if($clone_name eq "Empty");
  next if($clone_name eq "EMPTY");
  my $mtp_name = $feature_array[1];
  if(exists($comment_hash{$clone_name})) {
    $comment_hash{$clone_name} .= " $mtp_name";
  } # if
  else {
    $comment_hash{$clone_name} = $mtp_name;
  } # else
#   print ". ";
} # for

open(OUTPUT, ">", $output_str)
  or die "Cannot open output file:  $!\n";
  foreach my $each_clone (keys %comment_hash) {
    print OUTPUT "$each_clone\t$comment_hash{$each_clone}\n";
  } # foreach
close OUTPUT;

# display the time
my $run_time = Time::HiRes::time - $start_time;
if($run_time < 60) {
  print "\nRun time:  $run_time seconds\n";
} # if
else {
  my $minutes = int($run_time / 60);
  $run_time %= 60;
  print "\nRun time:  $minutes minutes, $run_time seconds\n";
} # if