#!/usr/local/bin/perl5 -w
#
# combine-scores [-sum|-min|-max] [-at] file.rdb file2.rdb ...
#
# merges several different rdb files that contain scores taking the
#	sum, minimum, or maximum of the scores.
# 
# The resultant rdb file is sent to stdout.
# If "-at" is specified with -min or -max, 
# then the output also has a column reporting which file produced the score.

$tmp = "tmp-combine-$$";

# create empty tmp file
open(TMP,">$tmp") || die "Can't write to $tmp\n";

$use_sum = 0;
$use_max=0;
$use_min=1;
$do_at = 0;
foreach $file (@ARGV)
{
    if ($file eq "-sum")
    {    $use_sum=1; $use_max=0; $use_min=0;
    }
    elsif ($file eq "-min")
    {	$use_sum = 0; $use_max=0; $use_min=1;
    }
    elsif ($file eq "-max")
    {	$use_sum = 0; $use_max=1; $use_min=0;
    }
    elsif ($file eq "-at")
    {	$do_at=1;
    }
    else
    {   $do_at = $do_at && !$use_sum;
        open(RDB_IN, "column TARGET HMM SCORE < $file | headchg -del |");
	while(<RDB_IN>)
	{    /^(\S+\t\S+)\t(\S+)$/ && ($key = $1) && ($score = $2);
	    if ($do_at)
	    {    print TMP "$key\t$score\t$file\n";
	    }
	    else
	    {    print TMP "$key\t$score\n";
	    }
	}
	close(RDB_IN);
    }
}    
close(TMP);

if (!$do_at)
{   print "TARGET\tHMM\tSCORE\n10S\t6S\t10N\n";
}
else
{   print "TARGET\tHMM\tSOURCE\tSCORE\n10S\t6S\t20S\t10N\n";
}

$old1 = "";
$old0 = "";
$sum=0;		$min=99999;	$max=-99999;
open(TMP, "sort $tmp|");
while(<TMP>)
{   chomp;
    @cols = split(/\t/);
    if ($cols[0] ne $old0 || $cols[1] ne $old1)
    {    # new key
	print_key() if ($old0 ne "");
	$sum=$cols[2];	$min=$sum;
	$old0=$cols[0]; $old1=$cols[1];
	next;
    }
    $sum += $cols[2] if $use_sum;
    if ($use_min && $min>$cols[2])
    {   $min = $cols[2];
	$at = $cols[3] if $do_at;
    }
    if ($use_max && $max<$cols[2])
    {   $max = $cols[2];
	$at = $cols[3] if $do_at;
    }
}
print_key();
close TMP;
unlink $tmp;

exit 0;

sub print_key()
{    
    if ($use_sum)
    {	print "$old0\t$old1\t$sum\n";
    }
    elsif ($do_at)
    {    print "$old0\t$old1\t$at\t$min\n";
    }
    else
    {    print "$old0\t$old1\t$min\n";
    }
}