#!/usr/bin/perl
# Description: This script runs the MaxAlign 1.1.ws0 Web Service. It requires no input; to be used for testing in the EMBRACE WS Registry.
# Author: Edita Bartaseviciute
# Email: edita@cbs.dtu.dk
# Version: 1.1 ws0
# Date: 2009-07-10
# usage: perl test_maxalign.pl 

# include standard XML::Compile helper functions (used to initiate WSDL proxys)
use strict;
require "xml-compile.pl";

# create proxy to maxalign
my $maxalign = WSDLclient ( 'http://www.cbs.dtu.dk/ws/MaxAlign/MaxAlign_1_1_ws0.wsdl' );


my @fasta;
my $entry = -1;
while (<DATA>) {
	chomp;
	if (/^>(.*)/) {
		my ($id , $comment) = split (" ",$1);
		$entry++;
		$fasta[$entry]->{id} = $id;
		$fasta[$entry]->{comment} = $comment if defined $comment;
	} else {
		$fasta[$entry]->{seq} .= $_;
	}
}
# Create 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 = $maxalign->{maxalign}->(
	 parameters => {
	  parameters => {
	   alignment => {sequence => [@sequence]} } });
# Print the response
print "# Original number of sequences: $response->{parameters}->{output}->{originalsequencenumber}\n";
print "# Original total number of columns: $response->{parameters}->{output}->{originalcolumnnumber}\n";
print "# Original ungapped columns: $response->{parameters}->{output}->{originalungapcolumnnumber}\n";
print "# Original alignment area: $response->{parameters}->{output}->{originalalignmentarea}\n\n";
if (exists $response->{parameters}->{output}->{resultsequencenumber} and $response->{parameters}->{output}->{resultungapcolumnnumber} and $response->{parameters}->{output}->{resultalignmentarea} ) {
	print "# Number of sequences in result: $response->{parameters}->{output}->{resultsequencenumber}\n";
	print "# Ungapped columns in result: $response->{parameters}->{output}->{resultungapcolumnnumber}\n";
	print "# Resulting alignment area: $response->{parameters}->{output}->{resultalignmentarea}\n\n";
	print "#New alignment in fasta format:\n\n";
	foreach my $seqref (@{$response->{parameters}->{output}->{resultalignment}->{sequence}}) {
   	print ">$seqref->{id}";
   	print exists $seqref->{comment} ? " $seqref->{comment}\n" : "\n";
   	for (my $i = 0; $i < length $seqref->{seq}; $i += 60) {
      	print substr($seqref->{seq}, $i, 60), "\n"; }
	}
}
else {
	print "This alignment can't be improved by removing sequences\n";
}
__DATA__
>Q9F0G5_STRMU/12-64 
TKTFDVQEDLFEFNITDTI-V--lQASDSPDThsKVGSFSICPP--R--KT-SvSFNSYCC
>Q9ETI4_STRMU/12-65 
TETFDVQEDLFAFDTTDTT-I--vASNDDPDT--RFSSLSLCSLGCTgvKN-P-SFNSYCC
>Q9FDN2_STRMU/12-64 
TKTFDVQEDLFEFNITDTI-VlqvSDSPGTHS--KVGSFSICPP--R--KT-SvSFNSYCC
>Q76HP8_9LACT/17-37 
--------------------------DSGAST--RITSISLCTPGC---KT-G--------
>Q2L5U7_CLOPE/4-33 
-----------DFDLNLKMET---GSSDDKEP--RITSVAYCTPGC---------------
>SPAS_BACSU/6-35 
-----------DFDLDVVKVS---KQDSKITP--QWKSESLCTPGC---------------
>Q5L3A9_GEOKA/22-34 
-------------------------------P--NITSKSLCTPGC---------------
>LANE_STAEP/1-52 
MEAVKEKNDLFNLDVKVNA-K--eSNDSGAEP--RIASKFICTPGCA--KT-G-SFNSYCC
>Q9RPL9_STRMU/12-64 
TKTFDVQEDLFEFNITDTI-V--lQASDSPDThsKVGSFNICPPRKI--SV---SFNSYCC
>Q2FXB2_STAAU/1-47 
MEKV------LDLDVQVKANN--nSNDSAGDE--RITSHSLCTPGCA--KT-G-SFNSFCC
>Q8VT57_STRMU/12-63 
TETFDVQENLFTFDTTDTI-V--aESNDDPDT--RFKSWSFCTPGCA--KT-G-SFNSYCC
>Q8NVW4_STAAW/1-47 
MEKV------LDLDVQVKGNN--nTNDSAGDE--RITSHLFCSFGCE--KT-G-SFNSFCC
>Q8VT56_STRMU/12-64 
SDTLDVREDLFDLNITDTA-V--iQSNDSPDTrsRVGSHSICPP--R--KTsA-SFNSYCC
>Q7B8R0_9LACT/17-37 
--------------------------DSGASP--RITSISLCTPGC---KT-G--------
>Q6G8B1_STAAS/1-47 
MEKV------LDLDVQVKANN--nSNDSAGDE--RITSHSLCTPGCA--KT-G-SFNSFCC
>O68587_STRMU/25-53 
TKTFDVQEDLFEFNITDTIVL---QASDSPDT-----------------------------
>Q45RP0_LACLA/17-37 
--------------------------DSGASP--RITSISLCTPGC---KT-G--------
>LANZ_LACLA/17-37 
--------------------------DSGASP--RITSISLCTPGC---KT-G--------
>Q2YTQ7_STAAB/1-47 
MENV------LDLDVQVKA-Kn-dTSDSAGDE--RITSFIGCTPGCG--KT-G-SFNSFCC
>Q2QBT0_STRUB/1-43 
-----MNNEDFNLDLIKIS-K--e-NNSGASP--RITSKSLCTPGC---KT-G-ILMT--C
>Q93GH5_BACSU/9-38 
-----------DFDLDVVKVS---KQDSKITP--QVLSKSLCTPGC---------------
>Q6G8B0_STAAS/1-47 
MEKV------LDLDVQVKGNN--nTNDSAGDE--RITSHLFCSFGCE--KT-G-SFNSFCC
>LANM_STRMU/1-22 
-----------------------------------FKSWSFCTPGCA--KT-G-SFNSYCC
>LANN_STRMU/12-63 
TETFDVQEDLFAFDTTDTT-I--vASNDDPDT--RFKSWSLCTPGCA--RT-G-SFNSYCC
>Q7BB86_9LACT/17-37 
--------------------------DSGASP--RITSISLCTPGC---KT-G--------
>Q7DH25_LACLA/17-37 
--------------------------DSGASP--RITSISLCTPGC---KT-G--------
>Q2YTQ5_STAAB/1-47 
MEKV------LDLDVQVKGNN--nTNDSAGDE--RITSHFLCSFGCG--KT-G-SFNSFCC
>LANG_STAGA/1-52 
MEAVKEKNELFDLDVKVNA-K--eSNDSGAEP--RIASKFLCTPGCA--KT-G-SFNSYCC
>LANN_LACLA/17-37 
--------------------------DSGASP--RITSISLCTPGC---KT-G--------
>Q8NVW5_STAAW/1-47 
MEKV------LDLDVQVKANN--nSNDSAGDE--RITSHSLCTPGCA--KT-G-SFNSFCC
>Q5HEV2_STAAC/1-47 
MEKV------LDLDVQVKANN--nSNDSAGDE--RITSHSLCTPGCA--KT-G-SFNSFCC
>Q1W155_BACSU/6-35 
-----------DFDLDVVKVS---KQDSKITP--QWKSESLCTPGC---------------
>Q2FFS0_STAA3/1-47 
MEKV------LDLDVQVKANN--nSNDSAGDE--RITSHSLCTPGCA--KT-G-SFNSFCC