#!/usr/bin/python
# Description: Use eFetch to retrieve a protein sequence given its id and predict if it's a signalling peptide
# Author: Francisco Roque
# Email: chico@cbs.dtu.dk
# Version: SignalP 3.1
# Date: 2008-01-23
# test with sequence Q9BS26

import sys, time
from SOAPpy import WSDL

# Receive the protein id from the command line
protid = sys.argv[1]

# Define wsdl locations
EFetchWSDL = 'http://www.ncbi.nlm.nih.gov/entrez/eutils/soap/efetch.wsdl'
SignalPWSDL = 'http://www.cbs.dtu.dk/ws/SignalP/SignalP_3_1.wsdl'

# Establish the endpoint for the eFetch wsdl 
EUtilsProxy = WSDL.Proxy(EFetchWSDL)

# Define the input data to eFetch
eFetchIn = { 'db' : 'protein',
			 'id' : protid}

# Run the run_eFetch service with the input data
eFetchOut = EUtilsProxy.run_eFetch(eFetchIn)

print 'Fetched sequence %s' % eFetchOut['GBSeq']['GBSeq_primary-accession']

# Define Signalp input data (a bit more complex)
SignalPIn = {'organism' : 'euk', 
			 'sequences' : { 'entry' : { 'ident' : eFetchOut['GBSeq']['GBSeq_primary-accession'],
										 'seq' : eFetchOut['GBSeq']['GBSeq_sequence'] }
			}}
			 

# Establish the endpoint for the SignalP wsdl 
SignalPProxy = WSDL.Proxy(SignalPWSDL)

# The service will return a jobid (Asynchronous service)
jobid = SignalPProxy.runService(SignalPIn)

print 'SignalP service launched with jobid: %s' % jobid.jobid

# The pollQueue service needs to be run to get the job status 
job = SignalPProxy.pollQueue(jobid)

# do not continue until the job is finished
while job.status != 'FINISHED':
	job = SignalPProxy.pollQueue(jobid)
	print 'The job is %s' % job.status
	time.sleep(5)

print 'The job is FINISHED, fetching results.'

# Getting the output of the server
SignalPOut = SignalPProxy.fetchResult(jobid)

# loop over the output and print if predicted as a signalling peptide
for record in SignalPOut.prediction.gff_record:
	if record.comment == 'Y':
		print "seqname: %s" % record.seqname
		print "score: %s" % record.score
		print "seq: %s" % eFetchOut['GBSeq']['GBSeq_sequence'].upper()
		print "start: %s" % record.start
		print "end: %s" % record.end