#!/usr/bin/perl -w

# C.Vogel, UT Austin, TX; April 2008

# input: 
# 	.wekaout files (saved resultbuffers) from WEKA predictions
# 	.names files in same directory as input .arff files (input for WEKA prediction) -- need to be in same order as predictions in .wekaout!!!
# output:
#	.oi file: protein_id *tab* Oi-value *tab* total_peptides
# notes:
# 	Be in directory with WEKAOUT files (cutandpaste from wekapredictions) and NAMES file (ID _tab_ Peptide)
#	Files should be numbered as _number.names or _number.wekaout

# ======================================================================================
use strict;
$|=1;

die "\nnp_PeptidePredictions_to_ProteinOi.pl <.names filename> <.wekaout filename> <number of files> <out file name>\nNOTES:\tBe in directory with WEKAOUT files (saved resultbuffer from WEKA prediction) and NAMES file (ID _tab_ Peptide)\n\tFiles should be numbered as _number.names or _number.wekaout\n" unless ($#ARGV==3);
my $namesfile = $ARGV[0]; # name of _[number].names files in directory
my $wekafile = $ARGV[1]; # name of _[number].wekaout files in directory (with WEKA resultbuffers)
my $number = $ARGV[2]; # number of files in directory (from 0 to X)
my $out = $ARGV[3]; # 

# file formats:
# .names file: 	(created by np_arf_to_arff_TEST.pl)
#			id *tab* peptide 
# .wekaout file: (saved resultbuffer from WEKA predictions)
# 			     1          ?        2:0      +   0.01  *0.99 


# ======================================================================================
# define file numbers (note: currently starting from 0):
my @file_numbers = ();
for (my $k=0; $k<=($number-1);$k++) {
	push(@file_numbers, $k);
} # for (... 

my %OIS; # hash{protein} = sum of probabilities (for all peptides)
my %PEPS; # hash(protein) = count peptides per protein


# ======================================================================================
# collect data and prting --- 
foreach my $numb ( @file_numbers )  {
	print "\# PARSING FILE NUMBER\t$numb\n";

	# read in peptides/IDs (in same order as below the prediction!!!) --- 
	my @STORE_NAMES = (); # store ID and peptide
	open (NAM, "$namesfile$numb.names") or die "Can't open $namesfile$numb.names\n";
	while ( my $line = <NAM> ) {
		next if length(chomp $line) == 0;
		my @a = split("\t", $line);
		die "ERROR STRANGE LINE in $namesfile$numb.names at $line\n" unless scalar( @a ) == 2;		
		push(@STORE_NAMES, $line); # # store content
	} # while
	close NAM;
	print "\# STORING\t", scalar( @STORE_NAMES ), "\tNAMES from $namesfile$numb.names\n";

	
	# go through all predictions, map to peptides, and sum predictions for each peptide ---
	my @PREDS = `grep '?' $wekafile$numb.wekaout`;
	print "\# PARSING \t", scalar( @PREDS ), "\tPREDICTIONS from $wekafile$numb.wekaout\n";
	
	# mapping and counting
	for ( my $k=0; $k<= $#PREDS; $k++ ) {
		# print "last line $k\n" if $k>199999;
		
		my ($id, $peptide) = split("\t", $STORE_NAMES[$k]);
		
		my $cline = $PREDS[$k];
		chomp $cline;
		$cline =~ s/\*//g;
		if ( $cline =~ /\+\s+(.*\d+.*)\s+(.*\d+.*)/ ) {
			my $pred = $1;
			$PEPS{$id}++;
			$OIS{$id} += $pred;
		} # if
		else {
			# print "ERROR $cline does not match pattern\n";
		} # else
		
	} #	for ( my $k=0; $k<= $#PREDS; $k++ ) {

} # foreach my $numb ( @file_numbers )  {  

### print out ----
`rm -f $out.oi`;
open (OUT, ">>$out.oi");

print "\# PRINTING Oi file $out.oi\n";
print OUT "\# PROTEIN_ID\tOi\tTOTAL_UNIQUE_PEPTIDES_EXPECTED\n";
foreach my $id ( keys( %OIS ) )  {
	print OUT "$id\t$OIS{$id}\t$PEPS{$id}\n";
} #foreach
 
close OUT;

# ======================================================================================
# ======================================================================================