#!/usr/bin/perl
# Rank array data in accordance with their P-values
# Script author: AR, 2007.
use Data::Dumper;

my $pvalues_file = $ARGV[0];
my $MAX_TOP_ENTRIES = 20;

#open PVAL, "< $pvalues_file" or die "Error: Can't open $pvalues_file: $!";

my %arrays;
my @profile_names;

while (my $line = <DATA>) {
    chomp $line;
    if ($line =~ /^ARRAYS/) {
        my $hdr;
        ( $hdr, @profile_names ) = split( ' ', $line );

    } else {
        ## Split first on the pipes, leaving the values attached to the end of the last field
        my( $array_name, @pvalues ) = split( '\|', $line ); 

        ## Then split the last field on whitespace and overlay the fields you which to discard
        ## Discarding the unwanted first two fields of the second split at the same time.
        @pvalues[ 1 .. 6 ] = ( split( '\s+', $pvalues[ 2 ] ) )[ 2 .. 7 ];

        for my $i ( 0 .. $#profile_names ) {
            $arrays{ $array_name }{ $profile_names[$i] } = $pvalues[$i];
        }
    }
}


print Dumper \%arrays;
exit;

foreach $array (keys %arrays) {
    my $top_count = 0;
    print "$array\t";
    %pvalues = %{$arrays{$array}};
        @profiles_sorted = sort { $pvalues{$a} <=> $pvalues{$b} } ( keys %pvalues );
    foreach $key (@profiles_sorted) {
        $top_count++;
            if ($top_count <= $MAX_TOP_ENTRIES) {
                  print "$key:$pvalues{$key}\t";
        }
    }

      print "\n";
}
__DATA__
ARRAYS    V1.4HFChip-1    V1.4HFChip-10    V1.4HFChip-100    V1.4HFChip-11    V1.4HFChip-12    V1.4HFChip-13
37133 | Tula virus | Bunyaviridae | V1.3_110017:22.1    0.539026    0.357762    0.801409    0.315076    0.207579    0.946322
263532 | Possum enterovirus W6 | Picornaviridae | V1.3_116027:82.1    0.242743  0.712059    0.474686    0.738211    0.26494    0.529945
271479 | Papaya leaf curl Guandong virus | Geminiviridae | V1.3_105649:75.8    0.291412    0.726736    0.277159    0.893388    0.24579    0.904211
12202 | Lettuce mosaic virus | Potyviridae | V1.3_118815:65.4    0.39146    0.567612    0.771404    0.671439    0.427434    0.855816
116056 | Pelargonium zonate spot virus | Bromoviridae | V1_111931:65.5    0.704965    0.750921    0.66365    0.835392    0.654149    0.04262
45709 | Sabia virus | Arenaviridae | V1_112261:16.8    0.392471    0.740175    0.584603    0.861441    0.434677    0.758832
130556 | Culex nigripalpus NPV | Baculoviridae | V1_112047:15.8    0.315955    0.882084    0.551393    0.909915    0.088346    0.745482
312349 | Procyon lotor papillomavirus type 1 | Papillomaviridae | V1.3_113827:83.8    0.652409    0.200222    0.65569    0.239118    0.537655    0.889673
243550 | Calicivirus isolate TCG | Caliciviridae | V1.3_115411:78.6    0.324359   0.820308    0.238306    0.88163    0.311354    0.741035
150285 | Garlic virus E | Flexiviridae | V1.3_103783:90.0    0.267302    0.809609    0.55432    0.908932    0.193653    0.718928