#!/usr/bin/perl -w
# Parse the GO ontology

#open(PO, "process.ontology.2002-06-01");
#open (PO, "process.ontology.2003-12-12.txt");
open(PO, "process.ontology");
# open(PO, "test.txt");
while (<PO>) {
	chomp;
	unless (/^!/) {	# Header
		# print "line = $_\n";
		
		# Remove leading spaces and start character
		# Need to count leading spaces
		/^\s*/;
		$space = $&;
		$level = length $space;
		# print "level = $level\n";
		s/^\s*(\$|<|%)//;

		@data = split /<|%|\$/; 
		@id = ();	# Reset
		
		for $data (@data) {

			# Remove leading space
			$data =~ s/^\s*//;
			# print "data = $data\n";

			# Identify terms
			@term = split /\s*;\s*/, $data;
			for $i (0 .. $#term) {
				# print "term = $term[$i]\n";
				if ($term[$i] =~ /^GO:[0-9]+/) {
					$description = $term[$i - 1];
					$id = $term[$i];
					$id =~ s/\s+//g;	# Remove all spaces
					push @id, $id;
					# print "id = $id\tdescription = $description\n";
					
					# IDs can me multiple, separate single IDs
					@sid = split /\s*,\s*/, $id;
					for $sid (@sid) {
						# print "sid = $sid\n";
						$description{$sid} = $description;
					}	
				} 
			}
		}
#		print "level = $level\tid array = @id\n";
		
		$node_level = $level;
		$node[$node_level] = $id[0];
#		print "node_level = $node_level\tnode = $node[$node_level]\n";
		
		# Compute edges
		
		if ($node_level > 0) {
		    @parent = split /\s*,\s*/, $node[$node_level - 1];
		    @child = split /\s*,\s*/, $node[$node_level];
#		    print "parent = @parent\n";
#		    print "child = @child\n";
		    
		    for $parent (@parent) {
			for $child (@child) {
			    $edge = $parent."->".$child;
			    $graph{$edge} = 1;
#			    print "edge = $edge\n";
			}
		    }
		}


		# All the ids followed the first one that are seperated by "%" or "<" are parents of the first id -- Ronald made a major mistake!!!
		if ($#id > 0) {
		    for $i (1 .. $#id) {

			# Compute edges

			@parent = split /\s*,\s*/, $id[$i];
			@child = split /\s*,\s*/, $id[0];
#			print "parent = @parent\n";
#			print "child = @child\n";
			
			for $parent (@parent) {
			    for $child (@child) {
				$edge = $parent."->".$child;
				$graph{$edge} = 1;
#				print "edge = $edge\n";
			    }
			}
		    }
		}
	    }	
    }
close(PO);

# Print node descriptions

open(TD, ">id_description_go_process_hy.txt");
open(PN, ">go_process_node_hy.txt");
@sid = sort(keys %description);
$count = 0;
for $sid (@sid) {
	print TD "$sid\t$description{$sid}\n";
	print PN "$count\t$sid\n";
	$count++;
}
close(TD);
close(PN);


# Print graph (edges)

open(GR, ">go_process_graph_hy.txt");
@edge = sort(keys %graph);
for $edge (@edge) {
	print GR "$edge\n";
}
close(GR);