#!/usr/bin/perl


use strict;
use constant SUCCESS => 0;
use constant ERROR => -1;
require XML::Simple;
use subs ;
use Data::Dumper;
use File::Basename;
use Benchmark;

# This script will take a XML file of probes, and look up the location in the human
# the location of the probes.

my $probe_file = getParams();


my $in1= new XML::Simple();
my $probes = $in1->XMLin("$probe_file");
#print Dumper \$probes;
#exit;
#open (CSV,">/home/bcarlson/data/Target3/all_probes.cvs");
#open (XML,">/home/bcarlson/data/Target3/all_probes.xml");
open (CSV,">all_probes.cvs");
open (XML,">all_probes.xml");
open (BLAST,">all_probes.blast");
print XML "<?xml version='1.0' encoding='ISO-8859-1'?>\n";
print XML "<candidates>\n"; 
for (my $i=0; $i<@{$probes->{probe}}; $i++)
{
  my $species = $probes->{probe}->[$i]->{species};
  my $target_probe = $probes->{probe}->[$i]->{sequence};
  my $hybridize_file = "human_".$species."_T5.verbose.probes.xml";
  my $in2 = new XML::Simple();
  my $results = $in2->XMLin("$hybridize_file");
  my $j=0;
  my $max = @{$results->{parent_alignment}};
  my $location = 0;
  while (($location == 0) && ($j<$max))
  {
    #print "Line $i : $target_probe $results->{parent_alignment}->[$i]->{probe}->{sequence}\n";
    if ($target_probe eq $results->{parent_alignment}->[$j]->{probe}->{sequence})
    {
      $location = $results->{parent_alignment}->[$j]->{species}->{species1}->{start};
      $location = $location + $results->{parent_alignment}->[$j]->{probe}->{offset};
    } 
    $j++;
  } 
  my $cat1_count = $probes->{probe}->[$i]->{cat1_count};
  my $cat1_score = $probes->{probe}->[$i]->{cat1_score};
  my $cat2_count = $probes->{probe}->[$i]->{cat2_count};
  my $cat2_score = $probes->{probe}->[$i]->{cat2_score};
  my $cat3_count = $probes->{probe}->[$i]->{cat3_count};
  my $cat3_score = $probes->{probe}->[$i]->{cat3_score};

  print BLAST ">$i $species $location $cat1_count $cat1_score $cat2_count $cat2_score $cat3_count $cat3_score\n";
  print BLAST "$target_probe\n\n";

  print CSV "$i,$cat1_score,$cat1_count,$cat2_score,$cat2_count,$cat3_score,$cat3_count,$location,$species,$target_probe\n";

  print XML "  <probe>\n";
  print XML "    <id>$i</id>\n";
  print XML "    <sequence>$target_probe</sequence>\n";
  print XML "    <species>$species</species>\n";
  print XML "    <location>$location</location>\n";
  print XML "    <cat1_count>$cat1_count</cat1_count>\n";
  print XML "    <cat1_score>$cat1_score</cat1_score>\n";
  print XML "    <cat2_count>$cat2_count</cat2_count>\n";
  print XML "    <cat2_score>$cat2_score</cat2_score>\n";
  print XML "    <cat3_count>$cat3_count</cat3_count>\n";
  print XML "    <cat3_score>$cat3_score</cat3_score>\n";
  print XML "  </probe>\n\n";
}
print XML "</candidates>\n"; 
#########################################################################
# getParams - gets all options from the command line or keeps defaults
#             dies if required options are not included
#
# uses Getopt::Std and reads @ARGV
# defaults can be modified here
# returns hash of options and values
# 
# THIS LIST IS OUT OF DATE... JUST LOOK BELOW
# pick_from_query - 0 to make overgo from reference sequence 
#                   1 to make from query
# identity_cutoff - minimum proportion identical bases for pip
# ref_offset      - offset from begining of reference sequence at which to
#                   begin looking for pips
# gap_length      - optimal spacing between overgos
# overgo_length   - the length of the overgo probe to design
# overgo_overlap  - the overlap between primers for the probe
# overgo_target_gc- the target GC percentage to go for
# overgo_gc_wiggle- the maximum difference from the overgo_target_gc 
#                   for an overgo
# debug           - write debugging messages to stdout and logfile
# overgo_filename - file to write picked overgos to
# all_overgo_filename - file to write all possible overgos to (defaults
#                       to none)
#########################################################################
sub getParams {
    my %opts;
    my $errstr = "";
    use Getopt::Long;
    use Data::Dumper;
 
    GetOptions(\%opts, 'file=s');
    if (!defined($opts{file}))
    {
      $errstr .= "ERROR : No probes sources defined.\n";
    }
    if ($errstr) {
        print <<"END_USAGE";
Description: 

  This script will take a XML file of probes, and look up the location in the human
  the location of the probes.

Usage: locate_bp [options] 

Probe Sources:
-file   <filename>  Hybridization file resultant file.

END_USAGE
        print "\n$errstr" if ($errstr);
        die("\n");
    }
    return $opts{file};
}