#!/usr/bin/perl
# Description: This script runs the NetChop 3.1.ws0 Web Service. It reads a FASTA file from STDIN and produces predictions.
# Author: Edita Bartaseviciute
# Email: edita@cbs.dtu.dk
# Version: 3.1 ws0
# Date: 2009-07-06
# usage: perl netchop.pl [-v integer] [-t float] < example.fsa


use strict;
# include standard XML::Compile helper functions (used to initiate WSDL proxys)
require "xml-compile.pl"; # downloadable from the same site as this script


#taking options from a command line
my ($method, $threshold);
for (my  $i = 0; $i  <= $#ARGV; $i ++) {
	if ($ARGV[$i] eq "-v") {
		$method = $ARGV[$i+1];
	}
	elsif ($ARGV[$i] eq "-t") {
		$threshold = $ARGV[$i+1];
	}
}

# create proxy to NetChop Web Service
my $netchop = WSDL2proxy ( 'http://www.cbs.dtu.dk/ws/NetChop/NetChop_3_1_ws0.wsdl' );

# append schema definitions
$netchop = appendSchemas ( $netchop , 
	"http://www.cbs.dtu.dk/ws/common/ws_common_1_0b.xsd" ,
	"http://www.cbs.dtu.dk/ws/NetChop/ws_netchop_3_1_ws0.xsd"
);
# create hash of operations from proxy
my %ops = addOperations ( $netchop ) ;

# Get sequence in fasta format from STDIN
my @fasta;
my $entry = -1;

while (<STDIN>) {
	if (/^>(.*)/) {
		my ($id , $comment) = split (" ",$1);
		$entry++;
		$fasta[$entry]->{id} = $id;
		$fasta[$entry]->{comment} = $comment if defined $comment;
	} elsif (/^([A-Za-z]+)/) {
		$fasta[$entry]->{seq} .= $1;
	}
}
# Create sequence for request
my @sequence;
for ( my $i = 0 ; $i < scalar ( @fasta ) ; $i ++ ) {
	push @sequence , { id => $fasta[$i]->{id} , comment => $fasta[$i]->{comment} , seq => $fasta[$i]->{seq} };
}
# Do the request
my $response;
if ((defined $method) and (defined $threshold)) {
	$response = $ops{runService}->(
	 parameters => {
	  parameters => {
	   method => $method ,
	   threshold => $threshold ,
	    sequencedata => {sequence => [@sequence]} } });
}
elsif ((defined $method) and !(defined $threshold)) {
	$response = $ops{runService}->(
	 parameters => {
	  parameters => {
	   method => $method , 
	     sequencedata => {sequence => [@sequence]} } });
	$threshold = 0.5; #default value, assigned here and used later for printing results
}
elsif (!(defined $method) and (defined $threshold)) {
	$response = $ops{runService}->(
	 parameters => {
	  parameters => {
	   threshold => $threshold ,
	    sequencedata => {sequence => [@sequence]} } });
	$method = 0; #default value, assigned here and used later for printing results
}
else {
	$response = $ops{runService}->(
	 parameters => {
	  parameters => {
	    sequencedata => {sequence => [@sequence]} } });
	$method = 0;
	$threshold = 0.5; #default values, assigned here and used later for printing results
}

# uncomment the two following lines to inspect the structure of $response
#use Data::Dumper;
#print Dumper($response);   

#get job id which can be used to get the results later
my $jobid;

if ( ! defined ( $response->{parameters}->{queueentry}) ) {
	die "error obtaining jobid\n";
} else {
	$jobid = $response->{parameters}->{queueentry}->{jobid};
	print STDERR "# waiting for job $jobid";
	my $status = "UNKNOWN";;
	# poll the queue
	while ( $status =~ /ACTIVE|RUNNING|QUEUED|WAITING|PENDING|UNKNOWN/ ) {
		my $response = $ops{pollQueue}->( job => { job  => { jobid => $jobid }   }) ;
		$status = $response->{queueentry}->{queueentry}->{status};
		print STDERR  ".";
	}
	die "\nunexpected job status '$status'\n" unless $status eq "FINISHED";
	print STDERR "\n# job has finished\n";
}
# when the job is done, fetch the result
$response = $ops{fetchResult}->(job => { jobid => $jobid });

# uncomment the two following lines to inspect the structure of $response
#use Data::Dumper;
#print Dumper($response); 

#printing the results (suitable for one sequence)
my $version_name;
if ($method == 0) {
	$version_name = 'C-term';
}
elsif ($method == 1) {
	$version_name = '20S';
}
print "NetChop-3.1 predictions using version $version_name. Threshold " . sprintf ("%.6f", $threshold) . "\n\n";

foreach my $ann (@{$response->{parameters}->{anndata}->{ann}}) {
	print "--------------------------------------\n pos  AA  C\t score",
	      "      Ident\n--------------------------------------\n";
	my $sequence = $sequence[0]->{seq};
	my $length = length ($sequence);
	my $no_cleavage = 0;
	my $no_aa = 0;
	
	foreach my $annrecord (@{$ann->{annrecords}->{annrecord}}) {
		my $pos = sprintf ("%4s", $annrecord->{pos});
		my $aa = substr($sequence, $pos-1, 1);
		my $comment;
		if (exists $annrecord->{comment}) {
			$comment = "S";
			$no_cleavage++;
		}
		else {
			$comment = ".";
		}
		my $score = sprintf ("%.6f", $annrecord->{score}[0]->{value});
				
		print "$pos   $aa  $comment   $score $ann->{sequence}->{id}\n";	
		
	}
print "--------------------------------------\n\n".
      "Number of cleavage sites $no_cleavage. Number of amino acids $length. Protein name $ann->{sequence}->{id}\n\n".
      "--------------------------------------\n";
}