#!/usr/bin/env perl

srand;
my $reiterate = 100000;
my ($zerocount, $correctcount, $wrongcount);

# Load in the list of snps and hla alleles from a file
my (%snphla, %guesslist, @guessfromhere, %count, %snphapcount, %guesscount, $snp, $hla, @hlas);
while (<>){
	last if ($_ eq "\n");
	($snp, $hla) = split;
	chomp ($snp);
	chomp ($hla);
#	print $snp, " ", $hla, "\n";
	$snphla{$snp}->{$hla} += 1 ;
	push (@guessfromhere, $snp);
	my $current = $guesslist{$snp};
	$guesslist{$snp} = "$current $hla";
	$countEntries ++;	
	$snphapcount{$snp} += 1;
}
  print $countEntries . "\n";



for ($i = 1; $i <= $reiterate ; $i++){

# Make the guess and return the array of values associated with it
	my $random = int(rand(scalar(@guessfromhere)));
	my $guess = $guessfromhere[$random];
	my @parts = split(" ", $guesslist{$guess}); 
	my $number = int(rand(scalar(@parts)));
	my $answer = @parts[$number];
#	print "guess " . $guess . " answer " . $answer .  "\n";

# Count the number of times a guess is guessed
	$guesscount{$guess} += 1;



# Drop it out! Remove the guess from the count in the snphla hash
	$snphla{$guess}->{$answer} -= 1 ;


# Move only the guess' counts into a sorted array 

	my @counts = values %{$snphla{$guess}};
	my @sorted_counts = sort { $b <=> $a } @counts;

	my $highest_count = shift @sorted_counts;
 	if ($highest_count != 0){

		my @matching_hla_types = ();

		foreach my $hla_type (keys %{$snphla{$guess}})
		{
		  if ($snphla{$guess}->{$hla_type} == $highest_count)
	 	 {
		    push(@matching_hla_types, $hla_type);
		  }
		}

		my $random_index = int(rand(scalar(@matching_hla_types)));

#		print "guess " . $guess . " count " . $highest_count . " hla " .
		  $matching_hla_types[$random_index] . "\n";
		if ($answer eq $matching_hla_types[$random_index]){
			$correctcount++;
#			print "$correctcount yeah!\n";
			} else {
			$wrongcount++;
#			print "nope!\n";
			}
		} else {
		$zerocount++;
#		print "zerocount!\n";
		}
# Add it back for the next round!
	$snphla{$guess}->{$answer} += 1 ;
}
print "The number of wrong guesses was: " . $wrongcount . "\n";
print "The number of correct guesses was: " . $correctcount . "\n";
print "The percentage right was: " .  $correctcount / ($reiterate - $zerocount) . "\n";
print "The number of singletons was: ". $zerocount . "\n";
foreach $key (sort keys %guesscount){
	$guessvalue = ($guesscount{$key})/$reiterate;
	$hapvalue = ($snphapcount{$key})/$countEntries;
	print "Haplotype " . $key . " guessed " . $guessvalue . " present " . $hapvalue . "\n";
}

exit;