#!/usr/bin/env python

from suds.client import Client
import os, sys
import cPickle as pickle

wsdlfile = "http://www.pdb.org/pdb/services/pdbws?wsdl"
client = Client(wsdlfile)

hot_pickle_filename = 'find_hot_structures.pickle'
if os.path.exists(hot_pickle_filename):
    hot_pickle = open(hot_pickle_filename)
    results = pickle.load(hot_pickle)
    hot_pickle.close()
else:
    results = client.service.getGenomeDetails()
    hot_pickle = open(hot_pickle_filename, 'w')
    pickle.dump(results, hot_pickle)
    hot_pickle.close()

pdb_gene_list = []
for result in results:
    pdb_gene_id = int(result.split()[2])
    if pdb_gene_id != 0:
        if pdb_gene_id not in pdb_gene_list:
            pdb_gene_list.append(pdb_gene_id)

pubmed_gene_list = []
pubmed_gene_pubs = []

for line in open('gene2pubmed_sorted_by_max_refs.out'):
    words = line.split()
    pubmed_gene_pub = int(words[0])
    pubmed_gene = int(words[1])

    pubmed_gene_pubs.append(pubmed_gene_pub)
    pubmed_gene_list.append(pubmed_gene)

no_structures = set(pubmed_gene_list) - set(pdb_gene_list)
#structures = set(pubmed_gene_list) & set(pdb_gene_list)

for count, gene, in zip(pubmed_gene_pubs, pubmed_gene_list):
    if count < 50:
        continue

    if gene in no_structures:
        print count, gene