#!/usr/bin/perl
# Description: This script runs the TargetP 1.1 ws0 Web Service. It reads FASTA from STDIN and produces predictions in a simple tables.
# Author: Edita Bartaseviciute
# Email: edita@cbs.dtu.dk
# Version: 1.1 ws0
# Date: 2009-10-07
# usage: perl targetp.pl -P|-N [-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 ($organism, $cutoff);
for (my $i = 0; $i <= $#ARGV; $i++) {
	if ($ARGV[$i] eq "-P" or $ARGV[$i] eq "-N") {
		$organism = substr ($ARGV[$i], 1, 1);
	}
	if ($ARGV[$i] eq "-t") {
		$cutoff = $ARGV[$i+1];
	}
}


# create proxy to TargetP Web Service
my $targetp = WSDL2proxy ( 'http://www.cbs.dtu.dk/ws/TargetP/TargetP_1_1_ws0.wsdl' );

# append schema definitions
$targetp = appendSchemas ( $targetp , 
	"http://www.cbs.dtu.dk/ws/common/ws_common_1_0b.xsd" ,
	"http://www.cbs.dtu.dk/ws/TargetP/ws_targetp_1_1_ws0.xsd"
);

# create hash of operations from proxy
my %ops = addOperations ( $targetp ) ;

# 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 $cutoff) {
	$response = $ops{runService}->(
	 parameters => {
	  parameters => {
	   organism => $organism,
	   mTP => $cutoff,
	   sequencedata => {sequence => [@sequence]} } });
}
else {
	$response = $ops{runService}->(
	 parameters => {
	  parameters => {
	   organism => $organism,
	    sequencedata => {sequence => [@sequence]} } });
}

# 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
my $version = substr($response->{parameters}->{anndata}->{annsource}->{version}, 0, 3);
print "### $response->{parameters}->{anndata}->{annsource}->{method} $version prediction results ##################################\n";
if ($organism eq "P") {
	print "Name\t\tLen\tcTP\tmTP\tSP\tother\tLoc\tRC\n",
		"----------------------------------------------------------------------\n"
}
elsif ($organism eq "N") {
	print "Name\t\tLen\tmTP\tSP\tother\tLoc\tRC\n",
		"----------------------------------------------------------------------\n"
}
	
foreach my $ann (@{$response->{parameters}->{anndata}->{ann}}) {
	my $seq;
	for ( my $i = 0 ; $i < scalar ( @sequence ) ; $i ++ ){
		if ($sequence[$i]->{id} eq $ann->{sequence}->{id}){
			$seq = $sequence[$i]->{seq};
		}
	}
	my $length = length ($seq);
	
	foreach my $annrecord (@{$ann->{annrecords}->{annrecord}}) {
		my $loc = substr ($annrecord->{comment}, 0, 1);
		print "$ann->{sequence}->{id}\t$length\t";
		if ($organism eq "P") {
			print "$annrecord->{score}[0]->{value}\t$annrecord->{score}[1]->{value}\t$annrecord->{score}[2]->{value}\t$annrecord->{score}[3]->{value}\t$annrecord->{comment}\t$annrecord->{score}[4]->{value}\n";
		}
		elsif ($organism eq "N") {
			print "$annrecord->{score}[0]->{value}\t$annrecord->{score}[1]->{value}\t$annrecord->{score}[2]->{value}\t$annrecord->{comment}\t$annrecord->{score}[3]->{value}\n";
		}
	}	
}
print "----------------------------------------------------------------------\n";