#!/usr/bin/perl -w

use strict;

# Goal  : 1. Format "go_process_graph.txt" into "process_graph.txt";
#         2. format "go_process_orfs.txt" into "process_terms_for_orf.txt";
# Author: Haiyuan Yu
# Date  : 04/19/2006

sub main;

#my $hostname = `hostname -s`;
my $hostname = `hostname`;
my $HOME = ($hostname =~ /^Legolas/ ) ? "/cygdrive/d/Work" : "/home/bh1/haiyuan/Linux_copy" ;

my $origin = "GO:0008150";

&main;

sub main {
    my (%GOf, %BPb, %BP_id); # %GOf: the forward edge of GO; %BPb: the backward edge of only BP in GO; %BP_id: all the go ids in BP;

    open (DATA, "go_process_graph_hy.txt") || die "Unable to open graph";
    open (OUT, ">process_graph_hy.txt");
    while (<DATA>) {
	chomp;
	my ($p, $c) = split(/\-\>/);
	$GOf{$p}{$c} = 1 if (($p ne "GO:0000004") && ($c ne "GO:0000004"));
    }
    close (DATA);
 
    print OUT "origin\-\>$origin";
    $BP_id{$origin} = 1;
    my %read = my @level = my @nlevel = ();
    push @level, $origin;
    while ($#level >= 0) {
	@nlevel = ();
	for my $aa (@level) {
	    if (!exists $read{$aa}) {
		$read{$aa} =1;
		for my $bb (keys %{$GOf{$aa}}) {
			push @nlevel, $bb;
			$BPb{$bb}{$aa} = 1;
			$BP_id{$bb} = 1;
			my $t = $aa ."_".$bb;
			print OUT " $aa\-\>$bb" if (!exists $read{$t});
			$read{$t} = 1;
		}
	    }
	}
	@level = ();
	push @level, @nlevel;
    }
    print OUT "\n";
    close (OUT);


    open (DATA, "go_process_orf.txt") || die "Unabl eto oepn orfs";
    open (OUT, ">process_terms_for_orf_hy.txt");
    while (<DATA>) {
	chomp;
	my ($orf, @goes) = split(/\s+/);
	my $key = 0; # to dermine whether an orf has any annotation in BP;
	for my $go (@goes) {
	    $key = 1 if (exists $BPb{$go});
	}
	if ($key) {
	    print OUT "$orf";
	    my $n = 0;
	    %read = @level = @nlevel = ();
	    for my $go (@goes) {
		if (exists $BPb{$go}) {
		    push @level, $go;
		    while ($#level >= 0) {
			@nlevel = ();
			for my $aa (@level) {
			    $n ++;
			    if ($n == 1) {
				print OUT "\t$aa" if (!exists $read{$aa});
				$read{$aa} = 1;
			    }
			    else {
				print OUT " $aa" if (!exists $read{$aa});
				$read{$aa} = 1;
			    }
			    for my $bb (sort keys %{$BPb{$aa}}) {
				push @nlevel, $bb if (!exists $read{$bb});
			    }
			}
			@level = ();
			push @level, @nlevel;
		    }
		}
	    }
	    print OUT "\n";
	}
    }
    close (DATA);
    close (OUT);
    
}