#!/usr/bin/python
# Description: Use eFetch to retrieve a protein sequence given it's id
# Author: Francisco Roque
# Email: chico@cbs.dtu.dk
# Version: eFetch 1.5b
# Date: 2008-01-23

import sys
from SOAPpy import WSDL


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

# Define wsdl locations
wsdlFile = 'http://www.ncbi.nlm.nih.gov/entrez/eutils/soap/efetch.wsdl'

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

# 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 the results
#print eFetchOut['GBSeq']['GBSeq_primary-accession']
#print eFetchOut['GBSeq']['GBSeq_sequence'].upper()