annotate data_manager/resource_building.py @ 28:808c8493ed4f draft

planemo upload commit 5c2c274361c0daceae1f678eca0e6c0d5b4ba4f7-dirty
author proteore
date Mon, 27 May 2019 07:49:19 -0400
parents 9a400ce8e4e6
children 9a40b72414de
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
1 # -*- coding: utf-8 -*-
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
2 """
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
3 The purpose of this script is to create source files from different databases to be used in other proteore tools
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
4 """
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
5
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
6 import os, sys, argparse, requests, time, csv, re, json, shutil, zipfile
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
7 from io import BytesIO
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
8 from zipfile import ZipFile
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
9 from galaxy.util.json import from_json_string, to_json_string
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
10
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
11 #######################################################################################################
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
12 # General functions
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
13 #######################################################################################################
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
14 def unzip(url, output_file):
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
15 """
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
16 Get a zip file content from a link and unzip
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
17 """
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
18 content = requests.get(url)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
19 zipfile = ZipFile(BytesIO(content.content))
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
20 output_content = ""
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
21 output_content += zipfile.open(zipfile.namelist()[0]).read()
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
22 output = open(output_file, "w")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
23 output.write(output_content)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
24 output.close()
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
25
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
26 def _add_data_table_entry(data_manager_dict, data_table_entry,data_table):
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
27 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {})
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
28 data_manager_dict['data_tables'][data_table] = data_manager_dict['data_tables'].get(data_table, [])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
29 data_manager_dict['data_tables'][data_table].append(data_table_entry)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
30 return data_manager_dict
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
31
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
32 #######################################################################################################
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
33 # 1. Human Protein Atlas
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
34 # - Normal tissue
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
35 # - Pathology
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
36 # - Full Atlas
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
37 #######################################################################################################
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
38 def HPA_sources(data_manager_dict, tissue, target_directory):
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
39 if tissue == "HPA_normal_tissue":
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
40 tissue_name = "HPA normal tissue"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
41 url = "https://www.proteinatlas.org/download/normal_tissue.tsv.zip"
1
0915249b8c4b planemo upload commit 4dd1a2f7d196a1d2e70fab379a2c08367da0fe94-dirty
proteore
parents: 0
diff changeset
42 table = "proteore_protein_atlas_normal_tissue"
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
43 elif tissue == "HPA_pathology":
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
44 tissue_name = "HPA pathology"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
45 url = "https://www.proteinatlas.org/download/pathology.tsv.zip"
1
0915249b8c4b planemo upload commit 4dd1a2f7d196a1d2e70fab379a2c08367da0fe94-dirty
proteore
parents: 0
diff changeset
46 table = "proteore_protein_atlas_tumor_tissue"
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
47 elif tissue == "HPA_full_atlas":
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
48 tissue_name = "HPA full atlas"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
49 url = "https://www.proteinatlas.org/download/proteinatlas.tsv.zip"
1
0915249b8c4b planemo upload commit 4dd1a2f7d196a1d2e70fab379a2c08367da0fe94-dirty
proteore
parents: 0
diff changeset
50 table = "proteore_protein_full_atlas"
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
51
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
52 output_file = tissue +"_"+ time.strftime("%d-%m-%Y") + ".tsv"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
53 path = os.path.join(target_directory, output_file)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
54 unzip(url, path) #download and save file
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
55 tissue_name = tissue_name + " " + time.strftime("%d/%m/%Y")
26
bf6940ff60a8 planemo upload commit 0290724216a2c445b4e28842153b84a1b28f4e9a-dirty
proteore
parents: 24
diff changeset
56 release = tissue_name.replace(" ","_").replace("/","-")
bf6940ff60a8 planemo upload commit 0290724216a2c445b4e28842153b84a1b28f4e9a-dirty
proteore
parents: 24
diff changeset
57 id = str(10000000000 - int(time.strftime("%Y%m%d")))
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
58
1
0915249b8c4b planemo upload commit 4dd1a2f7d196a1d2e70fab379a2c08367da0fe94-dirty
proteore
parents: 0
diff changeset
59
26
bf6940ff60a8 planemo upload commit 0290724216a2c445b4e28842153b84a1b28f4e9a-dirty
proteore
parents: 24
diff changeset
60 data_table_entry = dict(id=id, release=release, name = tissue_name, tissue = tissue, value = path)
1
0915249b8c4b planemo upload commit 4dd1a2f7d196a1d2e70fab379a2c08367da0fe94-dirty
proteore
parents: 0
diff changeset
61 _add_data_table_entry(data_manager_dict, data_table_entry, table)
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
62
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
63
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
64 #######################################################################################################
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
65 # 2. Peptide Atlas
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
66 #######################################################################################################
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
67 def peptide_atlas_sources(data_manager_dict, tissue, date, target_directory):
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
68 # Define organism_id (here Human) - to be upraded when other organism added to the project
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
69 organism_id = "2"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
70 # Extract sample_category_id and output filename
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
71 tissue=tissue.split(".")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
72 sample_category_id = tissue[0]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
73 tissue_name = tissue[1]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
74 output_file = tissue_name+"_"+date + ".tsv"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
75
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
76 query="https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id="+ \
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
77 sample_category_id+"&display_options=ShowAbundances&organism_id="+organism_id+ \
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
78 "&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf\
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
79 &QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
80
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
81 with requests.Session() as s:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
82 download = s.get(query)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
83 decoded_content = download.content.decode('utf-8')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
84 cr = csv.reader(decoded_content.splitlines(), delimiter='\t')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
85
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
86 uni_dict = build_dictionary(cr)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
87
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
88 #columns of data table peptide_atlas
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
89 tissue_id = tissue_name+"_"+date
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
90 name = tissue_id.replace("-","/").replace("_"," ")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
91 path = os.path.join(target_directory,output_file)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
92
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
93 with open(path,"w") as out :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
94 w = csv.writer(out,delimiter='\t')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
95 w.writerow(["Uniprot_AC","nb_obs"])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
96 w.writerows(uni_dict.items())
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
97
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
98 data_table_entry = dict(id=tissue_id, name=name, value = path, tissue = tissue_name)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
99 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_peptide_atlas")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
100
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
101 #function to count the number of observations by uniprot id
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
102 def build_dictionary (csv) :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
103 uni_dict = {}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
104 for line in csv :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
105 if "-" not in line[0] and check_uniprot_access(line[0]) :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
106 if line[0] in uni_dict :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
107 uni_dict[line[0]] += int(line[5])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
108 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
109 uni_dict[line[0]] = int(line[5])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
110
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
111 return uni_dict
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
112
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
113 #function to check if an id is an uniprot accession number : return True or False-
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
114 def check_uniprot_access (id) :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
115 uniprot_pattern = re.compile("[OPQ][0-9][A-Z0-9]{3}[0-9]|[A-NR-Z][0-9]([A-Z][A-Z0-9]{2}[0-9]){1,2}")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
116 if uniprot_pattern.match(id) :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
117 return True
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
118 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
119 return False
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
120
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
121 def check_entrez_geneid (id) :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
122 entrez_pattern = re.compile("[0-9]+|[A-Z]{1,2}_[0-9]+|[A-Z]{1,2}_[A-Z]{1,4}[0-9]+")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
123 if entrez_pattern.match(id) :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
124 return True
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
125 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
126 return False
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
127
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
128 #######################################################################################################
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
129 # 3. ID mapping file
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
130 #######################################################################################################
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
131 import ftplib, gzip
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
132 csv.field_size_limit(sys.maxsize) # to handle big files
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
133
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
134 def id_mapping_sources (data_manager_dict, species, target_directory) :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
135
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
136 human = species == "Human"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
137 species_dict = { "Human" : "HUMAN_9606", "Mouse" : "MOUSE_10090", "Rat" : "RAT_10116" }
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
138 files=["idmapping_selected.tab.gz","idmapping.dat.gz"]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
139
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
140 #header
28
808c8493ed4f planemo upload commit 5c2c274361c0daceae1f678eca0e6c0d5b4ba4f7-dirty
proteore
parents: 27
diff changeset
141 if human : tab = [["UniProt-AC","UniProt-ID","GeneID","RefSeq","GI","PDB","GO","PIR","MIM","UniGene","Ensembl_Gene","Ensembl_Transcript","Ensembl_Protein","neXtProt","BioGrid","STRING","KEGG"]]
808c8493ed4f planemo upload commit 5c2c274361c0daceae1f678eca0e6c0d5b4ba4f7-dirty
proteore
parents: 27
diff changeset
142 else : tab = [["UniProt-AC","UniProt-ID","GeneID","RefSeq","GI","PDB","GO","PIR","MIM","UniGene","Ensembl_Gene","Ensembl_Transcript","Ensembl_Protein","BioGrid","STRING","KEGG"]]
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
143
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
144 #get selected.tab and keep only ids of interest
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
145 selected_tab_file=species_dict[species]+"_"+files[0]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
146 tab_path = download_from_uniprot_ftp(selected_tab_file,target_directory)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
147 with gzip.open(tab_path,"rt") as select :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
148 tab_reader = csv.reader(select,delimiter="\t")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
149 for line in tab_reader :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
150 tab.append([line[i] for i in [0,1,2,3,4,5,6,11,13,14,18,19,20]])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
151 os.remove(tab_path)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
152
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
153 #print("selected_tab ok")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
154
20
29cf75c83618 planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 16
diff changeset
155 #get uniprot-AC reviewed
22
4296fb613d9d planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 21
diff changeset
156 organism = species_dict[species].split("_")[1]
4296fb613d9d planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 21
diff changeset
157 query = "https://www.uniprot.org/uniprot/?query=reviewed:yes+AND+organism:"+organism+"&format=list"
20
29cf75c83618 planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 16
diff changeset
158
29cf75c83618 planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 16
diff changeset
159 with requests.Session() as s:
29cf75c83618 planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 16
diff changeset
160 download = s.get(query)
29cf75c83618 planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 16
diff changeset
161 decoded_content = download.content.decode('utf-8')
29cf75c83618 planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 16
diff changeset
162 uniprot_reviewed_list = decoded_content.splitlines()
29cf75c83618 planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 16
diff changeset
163
29cf75c83618 planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 16
diff changeset
164 for line in tab[1:]:
29cf75c83618 planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 16
diff changeset
165 UniProtAC = line[0]
28
808c8493ed4f planemo upload commit 5c2c274361c0daceae1f678eca0e6c0d5b4ba4f7-dirty
proteore
parents: 27
diff changeset
166 if UniProtAC not in uniprot_reviewed_list :
808c8493ed4f planemo upload commit 5c2c274361c0daceae1f678eca0e6c0d5b4ba4f7-dirty
proteore
parents: 27
diff changeset
167 line[0]="NA"
20
29cf75c83618 planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 16
diff changeset
168
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
169 """
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
170 Supplementary ID to get from HUMAN_9606_idmapping.dat :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
171 -NextProt,BioGrid,STRING,KEGG
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
172 """
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
173
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
174 #there's more id type for human
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
175 if human : ids = ['neXtProt','BioGrid','STRING','KEGG' ] #ids to get from dat_file
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
176 else : ids = ['BioGrid','STRING','KEGG' ]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
177 unidict = {}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
178
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
179 #keep only ids of interest in dictionaries
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
180 dat_file=species_dict[species]+"_"+files[1]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
181 dat_path = download_from_uniprot_ftp(dat_file,target_directory)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
182 with gzip.open(dat_path,"rt") as dat :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
183 dat_reader = csv.reader(dat,delimiter="\t")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
184 for line in dat_reader :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
185 uniprotID=line[0] #UniProtID as key
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
186 id_type=line[1] #ID type of corresponding id, key of sub-dictionnary
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
187 cor_id=line[2] #corresponding id
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
188 if "-" not in id_type : #we don't keep isoform
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
189 if id_type in ids and uniprotID in unidict :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
190 if id_type in unidict[uniprotID] :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
191 unidict[uniprotID][id_type]= ";".join([unidict[uniprotID][id_type],cor_id]) #if there is already a value in the dictionnary
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
192 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
193 unidict[uniprotID].update({ id_type : cor_id })
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
194 elif id_type in ids :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
195 unidict[uniprotID]={id_type : cor_id}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
196 os.remove(dat_path)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
197
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
198 #print("dat_file ok")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
199
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
200 #add ids from idmapping.dat to the final tab
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
201 for line in tab[1:] :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
202 uniprotID=line[0]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
203 if human :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
204 if uniprotID in unidict :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
205 nextprot = access_dictionary(unidict,uniprotID,'neXtProt')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
206 if nextprot != '' : nextprot = clean_nextprot_id(nextprot,line[0])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
207 line.extend([nextprot,access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'),
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
208 access_dictionary(unidict,uniprotID,'KEGG')])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
209 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
210 line.extend(["","","",""])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
211 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
212 if uniprotID in unidict :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
213 line.extend([access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'),
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
214 access_dictionary(unidict,uniprotID,'KEGG')])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
215 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
216 line.extend(["","",""])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
217
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
218 #print ("tab ok")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
219
20
29cf75c83618 planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 16
diff changeset
220 #add missing nextprot ID for human or replace old ones
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
221 if human :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
222 #build next_dict
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
223 nextprot_ids = id_list_from_nextprot_ftp("nextprot_ac_list_all.txt",target_directory)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
224 next_dict = {}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
225 for nextid in nextprot_ids :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
226 next_dict[nextid.replace("NX_","")] = nextid
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
227 os.remove(os.path.join(target_directory,"nextprot_ac_list_all.txt"))
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
228
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
229 #add missing nextprot ID
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
230 for line in tab[1:] :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
231 uniprotID=line[0]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
232 nextprotID=line[13]
20
29cf75c83618 planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 16
diff changeset
233 if uniprotID in next_dict and (nextprotID == '' or (nextprotID != "NX_"+uniprotID and next_dict[uniprotID] == "NX_"+uniprotID)) :
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
234 line[13]=next_dict[uniprotID]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
235
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
236 output_file = species+"_id_mapping_"+ time.strftime("%d-%m-%Y") + ".tsv"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
237 path = os.path.join(target_directory,output_file)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
238
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
239 with open(path,"w") as out :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
240 w = csv.writer(out,delimiter='\t')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
241 w.writerows(tab)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
242
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
243 name_dict={"Human" : "Homo sapiens", "Mouse" : "Mus musculus", "Rat" : "Rattus norvegicus"}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
244 name = species +" (" + name_dict[species]+" "+time.strftime("%d/%m/%Y")+")"
21
026177e4ff4b planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 20
diff changeset
245 release = species+"_id_mapping_"+ time.strftime("%d-%m-%Y")
23
ebdd4961c6c2 planemo upload commit 03015f58ac7e7ba3cc44ba0d8899eacfaaf5a134-dirty
proteore
parents: 22
diff changeset
246 id = str(10000000000 - int(time.strftime("%Y%m%d"))) #new ids must be inferior to previous id -> sort by <filter> in xml only in descending order
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
247
21
026177e4ff4b planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents: 20
diff changeset
248 data_table_entry = dict(id=id, release=release , name = name, species = species, value = path)
2
2e34ee6d2d37 planemo upload commit 4dd1a2f7d196a1d2e70fab379a2c08367da0fe94-dirty
proteore
parents: 1
diff changeset
249 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_id_mapping_"+species)
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
250
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
251 def download_from_uniprot_ftp(file,target_directory) :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
252 ftp_dir = "pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
253 path = os.path.join(target_directory, file)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
254 ftp = ftplib.FTP("ftp.uniprot.org")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
255 ftp.login("anonymous", "anonymous")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
256 ftp.cwd(ftp_dir)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
257 ftp.retrbinary("RETR " + file, open(path, 'wb').write)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
258 ftp.quit()
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
259 return (path)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
260
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
261 def id_list_from_nextprot_ftp(file,target_directory) :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
262 ftp_dir = "pub/current_release/ac_lists/"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
263 path = os.path.join(target_directory, file)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
264 ftp = ftplib.FTP("ftp.nextprot.org")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
265 ftp.login("anonymous", "anonymous")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
266 ftp.cwd(ftp_dir)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
267 ftp.retrbinary("RETR " + file, open(path, 'wb').write)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
268 ftp.quit()
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
269 with open(path,'r') as nextprot_ids :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
270 nextprot_ids = nextprot_ids.read().splitlines()
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
271 return (nextprot_ids)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
272
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
273 #return '' if there's no value in a dictionary, avoid error
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
274 def access_dictionary (dico,key1,key2) :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
275 if key1 in dico :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
276 if key2 in dico[key1] :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
277 return (dico[key1][key2])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
278 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
279 return ("")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
280 #print (key2,"not in ",dico,"[",key1,"]")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
281 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
282 return ('')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
283
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
284 #if there are several nextprot ID for one uniprotID, return the uniprot like ID
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
285 def clean_nextprot_id (next_id,uniprotAc) :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
286 if len(next_id.split(";")) > 1 :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
287 tmp = next_id.split(";")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
288 if "NX_"+uniprotAc in tmp :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
289 return ("NX_"+uniprotAc)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
290 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
291 return (tmp[1])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
292 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
293 return (next_id)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
294
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
295
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
296 #######################################################################################################
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
297 # 4. Build protein interaction maps files
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
298 #######################################################################################################
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
299
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
300 def get_interactant_name(line,dico):
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
301
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
302 if line[0] in dico :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
303 interactant_A = dico[line[0]]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
304 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
305 interactant_A = "NA"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
306
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
307 if line[1] in dico :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
308 interactant_B = dico[line[1]]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
309 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
310 interactant_B = "NA"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
311
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
312 return interactant_A, interactant_B
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
313
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
314 def PPI_ref_files(data_manager_dict, species, interactome, target_directory):
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
315
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
316 species_dict={'Human':'Homo sapiens',"Mouse":"Mus musculus","Rat":"Rattus norvegicus"}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
317
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
318 ##BioGRID
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
319 if interactome=="biogrid":
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
320
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
321 tab2_link="https://downloads.thebiogrid.org/Download/BioGRID/Release-Archive/BIOGRID-3.5.167/BIOGRID-ORGANISM-3.5.167.tab2.zip"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
322
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
323 #download zip file
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
324 r = requests.get(tab2_link)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
325 with open("BioGRID.zip", "wb") as code:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
326 code.write(r.content)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
327
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
328 #unzip files
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
329 with zipfile.ZipFile("BioGRID.zip", 'r') as zip_ref:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
330 if not os.path.exists("tmp_BioGRID"): os.makedirs("tmp_BioGRID")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
331 zip_ref.extractall("tmp_BioGRID")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
332
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
333 #import file of interest and build dictionary
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
334 file_path="tmp_BioGRID/BIOGRID-ORGANISM-"+species_dict[species].replace(" ","_")+"-3.5.167.tab2.txt"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
335 with open(file_path,"r") as handle :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
336 tab_file = csv.reader(handle,delimiter="\t")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
337 dico_network = {}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
338 GeneID_index=1
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
339 network_cols=[1,2,7,8,11,12,14,18,20]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
340 for line in tab_file :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
341 if line[GeneID_index] not in dico_network:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
342 dico_network[line[GeneID_index]]=[[line[i] for i in network_cols]]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
343 else:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
344 dico_network[line[GeneID_index]].append([line[i] for i in network_cols])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
345
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
346 #delete tmp_BioGRID directory
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
347 os.remove("BioGRID.zip")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
348 shutil.rmtree("tmp_BioGRID", ignore_errors=True)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
349
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
350 #download NCBI2Reactome.txt file and build dictionary
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
351 with requests.Session() as s:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
352 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
353 r.encoding ="utf-8"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
354 tab_file = csv.reader(r.content.splitlines(), delimiter='\t')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
355
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
356 dico_nodes = {}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
357 geneid_index=0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
358 pathway_description_index=3
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
359 species_index=5
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
360 for line in tab_file :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
361 if line[species_index]==species_dict[species]:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
362 if line[geneid_index] in dico_nodes :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
363 dico_nodes[line[geneid_index]].append(line[pathway_description_index])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
364 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
365 dico_nodes[line[geneid_index]] = [line[pathway_description_index]]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
366
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
367 dico={}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
368 dico['network']=dico_network
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
369 dico['nodes']=dico_nodes
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
370
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
371 ##Bioplex
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
372 elif interactome=="bioplex":
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
373
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
374 with requests.Session() as s:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
375 r = s.get('http://bioplex.hms.harvard.edu/data/BioPlex_interactionList_v4a.tsv')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
376 r = r.content.decode('utf-8')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
377 bioplex = csv.reader(r.splitlines(), delimiter='\t')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
378
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
379 dico_network = {}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
380 dico_network["GeneID"]={}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
381 network_geneid_cols=[0,1,4,5,8]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
382 dico_network["UniProt-AC"]={}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
383 network_uniprot_cols=[2,3,4,5,8]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
384 dico_GeneID_to_UniProt = {}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
385 for line in bioplex :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
386 if line[0] not in dico_network["GeneID"]:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
387 dico_network["GeneID"][line[0]]=[[line[i] for i in network_geneid_cols]]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
388 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
389 dico_network["GeneID"][line[0]].append([line[i] for i in network_geneid_cols])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
390 if line[1] not in dico_network["UniProt-AC"]:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
391 dico_network["UniProt-AC"][line[2]]=[[line[i] for i in network_uniprot_cols]]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
392 else:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
393 dico_network["UniProt-AC"][line[2]].append([line[i] for i in network_uniprot_cols])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
394 dico_GeneID_to_UniProt[line[0]]=line[2]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
395
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
396 with requests.Session() as s:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
397 r = s.get('https://reactome.org/download/current/UniProt2Reactome.txt')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
398 r.encoding ="utf-8"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
399 tab_file = csv.reader(r.content.splitlines(), delimiter='\t')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
400
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
401 dico_nodes_uniprot = {}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
402 uniProt_index=0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
403 pathway_description_index=3
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
404 species_index=5
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
405 for line in tab_file :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
406 if line[species_index]==species_dict[species]:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
407 if line[uniProt_index] in dico_nodes_uniprot :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
408 dico_nodes_uniprot[line[uniProt_index]].append(line[pathway_description_index])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
409 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
410 dico_nodes_uniprot[line[uniProt_index]] = [line[pathway_description_index]]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
411
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
412 with requests.Session() as s:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
413 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
414 r.encoding ="utf-8"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
415 tab_file = csv.reader(r.content.splitlines(), delimiter='\t')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
416
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
417 dico_nodes_geneid = {}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
418 geneid_index=0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
419 pathway_description_index=3
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
420 species_index=5
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
421 for line in tab_file :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
422 if line[species_index]==species_dict[species]:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
423 if line[geneid_index] in dico_nodes_geneid :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
424 dico_nodes_geneid[line[geneid_index]].append(line[pathway_description_index])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
425 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
426 dico_nodes_geneid[line[geneid_index]] = [line[pathway_description_index]]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
427
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
428 dico={}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
429 dico_nodes={}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
430 dico_nodes['GeneID']=dico_nodes_geneid
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
431 dico_nodes['UniProt-AC']=dico_nodes_uniprot
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
432 dico['network']=dico_network
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
433 dico['nodes']=dico_nodes
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
434 dico['convert']=dico_GeneID_to_UniProt
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
435
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
436 ##Humap
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
437 elif interactome=="humap":
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
438
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
439 with requests.Session() as s:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
440 r = s.get('http://proteincomplexes.org/static/downloads/nodeTable.txt')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
441 r = r.content.decode('utf-8')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
442 humap_nodes = csv.reader(r.splitlines(), delimiter=',')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
443
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
444 dico_geneid_to_gene_name={}
9
cdd29444e0af planemo upload commit 71363136045353f422ff98219c1eb84f6fc6193a-dirty
proteore
parents: 8
diff changeset
445 dico_protein_name={}
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
446 for line in humap_nodes :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
447 if check_entrez_geneid(line[4]):
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
448 if line[4] not in dico_geneid_to_gene_name:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
449 dico_geneid_to_gene_name[line[4]]=line[3]
9
cdd29444e0af planemo upload commit 71363136045353f422ff98219c1eb84f6fc6193a-dirty
proteore
parents: 8
diff changeset
450 if line[4] not in dico_protein_name:
cdd29444e0af planemo upload commit 71363136045353f422ff98219c1eb84f6fc6193a-dirty
proteore
parents: 8
diff changeset
451 dico_protein_name[line[4]]=line[5]
cdd29444e0af planemo upload commit 71363136045353f422ff98219c1eb84f6fc6193a-dirty
proteore
parents: 8
diff changeset
452
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
453 with requests.Session() as s:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
454 r = s.get('http://proteincomplexes.org/static/downloads/pairsWprob.txt')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
455 r = r.content.decode('utf-8')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
456 humap = csv.reader(r.splitlines(), delimiter='\t')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
457
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
458 dico_network = {}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
459 for line in humap :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
460 if check_entrez_geneid(line[0]) and check_entrez_geneid(line[1]):
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
461
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
462 interactant_A, interactant_B = get_interactant_name(line,dico_geneid_to_gene_name)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
463
4
ce01295cd601 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 3
diff changeset
464 #first interactant (first column)
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
465 if line[0] not in dico_network:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
466 dico_network[line[0]]=[line[:2]+[interactant_A,interactant_B,line[2]]]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
467 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
468 dico_network[line[0]].append(line[:2]+[interactant_A,interactant_B,line[2]])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
469
4
ce01295cd601 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 3
diff changeset
470 #second interactant (second column)
ce01295cd601 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 3
diff changeset
471 if line[1] not in dico_network:
5
429e7481c392 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 4
diff changeset
472 dico_network[line[1]]=[[line[1],line[0],interactant_B,interactant_A,line[2]]]
4
ce01295cd601 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 3
diff changeset
473 else :
5
429e7481c392 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 4
diff changeset
474 dico_network[line[1]].append([line[1],line[0],interactant_B,interactant_A,line[2]])
4
ce01295cd601 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 3
diff changeset
475
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
476 with requests.Session() as s:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
477 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
478 r.encoding ="utf-8"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
479 tab_file = csv.reader(r.content.splitlines(), delimiter='\t')
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
480
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
481 dico_nodes = {}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
482 geneid_index=0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
483 pathway_description_index=3
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
484 species_index=5
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
485 for line in tab_file :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
486 if line[species_index]==species_dict[species]:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
487 #Fill dictionary with pathways
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
488 if line[geneid_index] in dico_nodes :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
489 dico_nodes[line[geneid_index]].append(line[pathway_description_index])
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
490 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
491 dico_nodes[line[geneid_index]] = [line[pathway_description_index]]
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
492
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
493 dico={}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
494 dico['network']=dico_network
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
495 dico['nodes']=dico_nodes
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
496 dico['gene_name']=dico_geneid_to_gene_name
9
cdd29444e0af planemo upload commit 71363136045353f422ff98219c1eb84f6fc6193a-dirty
proteore
parents: 8
diff changeset
497 dico['protein_name']=dico_protein_name
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
498
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
499 #writing output
27
9a400ce8e4e6 planemo upload commit bb113d19d3a756f70784e6a1433902888686ed96-dirty
proteore
parents: 26
diff changeset
500 output_file = species+'_'+interactome+'_'+ time.strftime("%Y-%m-%d") + ".json"
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
501 path = os.path.join(target_directory,output_file)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
502 name = species+" ("+species_dict[species]+") "+time.strftime("%d/%m/%Y")
27
9a400ce8e4e6 planemo upload commit bb113d19d3a756f70784e6a1433902888686ed96-dirty
proteore
parents: 26
diff changeset
503 release = species+"_"+interactome+"_"+ time.strftime("%Y-%m-%d")
26
bf6940ff60a8 planemo upload commit 0290724216a2c445b4e28842153b84a1b28f4e9a-dirty
proteore
parents: 24
diff changeset
504 id = str(10000000000 - int(time.strftime("%Y%m%d")))
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
505
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
506 with open(path, 'w') as handle:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
507 json.dump(dico, handle, sort_keys=True)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
508
26
bf6940ff60a8 planemo upload commit 0290724216a2c445b4e28842153b84a1b28f4e9a-dirty
proteore
parents: 24
diff changeset
509 data_table_entry = dict(id=id, release=release, name = name, species = species, value = path)
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
510 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_"+interactome+"_dictionaries")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
511
6
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
512 #######################################################################################################
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
513 # 5. nextprot (add protein features)
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
514 #######################################################################################################
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
515
8
d5badf9de1b0 planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents: 6
diff changeset
516 def Build_nextprot_ref_file(data_manager_dict,target_directory):
6
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
517 nextprot_ids_file = "nextprot_ac_list_all.txt"
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
518 ids = id_list_from_nextprot_ftp(nextprot_ids_file,target_directory)
12
f6afaa1f562c planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents: 11
diff changeset
519
f6afaa1f562c planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents: 11
diff changeset
520 output_file = 'nextprot_ref_'+ time.strftime("%d-%m-%Y") + ".tsv"
f6afaa1f562c planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents: 11
diff changeset
521 path = os.path.join(target_directory,output_file)
f6afaa1f562c planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents: 11
diff changeset
522 name = "neXtProt release "+time.strftime("%d-%m-%Y")
16
f75c525e0a4a planemo upload commit 04868e380c43447fac1309fc292785d67863a87b-dirty
proteore
parents: 15
diff changeset
523 release_id = "nextprot_ref_"+time.strftime("%d-%m-%Y")
12
f6afaa1f562c planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents: 11
diff changeset
524
13
098693479a9d planemo upload commit 64366857c1f29dd79a77dc331eb0153cb94cdb11
proteore
parents: 12
diff changeset
525 output = open(path, 'w')
098693479a9d planemo upload commit 64366857c1f29dd79a77dc331eb0153cb94cdb11
proteore
parents: 12
diff changeset
526 writer = csv.writer(output,delimiter="\t")
12
f6afaa1f562c planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents: 11
diff changeset
527
6
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
528 nextprot_file=[["NextprotID","MW","SeqLength","IsoPoint","Chr","SubcellLocations","Diseases","TMDomains","ProteinExistence"]]
14
3f9ba522bfe8 planemo upload commit 272d182247f4fa55a928ec8cdcb83eab50c7f6ef
proteore
parents: 13
diff changeset
529 writer.writerows(nextprot_file)
3f9ba522bfe8 planemo upload commit 272d182247f4fa55a928ec8cdcb83eab50c7f6ef
proteore
parents: 13
diff changeset
530
6
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
531 for id in ids :
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
532 #print (id)
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
533 query="https://api.nextprot.org/entry/"+id+".json"
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
534 resp = requests.get(url=query)
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
535 data = resp.json()
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
536
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
537 #get info from json dictionary
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
538 mass_mol = data["entry"]["isoforms"][0]["massAsString"]
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
539 seq_length = data['entry']["isoforms"][0]["sequenceLength"]
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
540 iso_elec_point = data['entry']["isoforms"][0]["isoelectricPointAsString"]
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
541 chr_loc = data['entry']["chromosomalLocations"][0]["chromosome"]
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
542 protein_existence = "PE"+str(data['entry']["overview"]['proteinExistence']['level'])
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
543
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
544 #put all subcell loc in a set
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
545 if "subcellular-location" in data['entry']["annotationsByCategory"].keys() :
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
546 subcell_locs = data['entry']["annotationsByCategory"]["subcellular-location"]
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
547 all_subcell_locs = set()
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
548 for loc in subcell_locs :
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
549 all_subcell_locs.add(loc['cvTermName'])
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
550 all_subcell_locs.discard("")
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
551 all_subcell_locs = ";".join(all_subcell_locs)
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
552 else :
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
553 all_subcell_locs = "NA"
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
554
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
555 #put all subcell loc in a set
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
556 if ('disease') in data['entry']['annotationsByCategory'].keys() :
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
557 diseases = data['entry']['annotationsByCategory']['disease']
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
558 all_diseases = set()
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
559 for disease in diseases :
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
560 if (disease['cvTermName'] is not None and disease['cvTermName'] != ""):
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
561 all_diseases.add(disease['cvTermName'])
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
562 if len(all_diseases) > 0 : all_diseases = ";".join(all_diseases)
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
563 else : all_diseases="NA"
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
564 else :
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
565 all_diseases="NA"
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
566
11
ac2cd728c40e planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 9
diff changeset
567 #get all tm domain
6
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
568 nb_domains = 0
11
ac2cd728c40e planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 9
diff changeset
569 if "transmembrane-region" in data['entry']['annotationsByCategory'].keys():
ac2cd728c40e planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 9
diff changeset
570 tm_domains = data['entry']['annotationsByCategory']["transmembrane-region"]
ac2cd728c40e planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 9
diff changeset
571 all_tm_domains = set()
ac2cd728c40e planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 9
diff changeset
572 for tm in tm_domains :
ac2cd728c40e planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 9
diff changeset
573 all_tm_domains.add(tm['cvTermName'])
ac2cd728c40e planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 9
diff changeset
574 nb_domains+=1
24
6cbb76823b7b planemo upload commit 3e8aad84b7fe9218370364b1aef3ea5779475f9f-dirty
proteore
parents: 23
diff changeset
575 #print "nb domains ++"
6cbb76823b7b planemo upload commit 3e8aad84b7fe9218370364b1aef3ea5779475f9f-dirty
proteore
parents: 23
diff changeset
576 #print (nb_domains)
15
5f39d3ef5fe3 planemo upload commit 45108c8cbeeccab12031bbf6e65f1177222d4e10
proteore
parents: 14
diff changeset
577 nextprot_file[:] = []
12
f6afaa1f562c planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents: 11
diff changeset
578 nextprot_file.append([id,mass_mol,str(seq_length),iso_elec_point,chr_loc,all_subcell_locs,all_diseases,str(nb_domains),protein_existence])
6
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
579 writer.writerows(nextprot_file)
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
580
24
6cbb76823b7b planemo upload commit 3e8aad84b7fe9218370364b1aef3ea5779475f9f-dirty
proteore
parents: 23
diff changeset
581 id = str(10000000000 - int(time.strftime("%Y%m%d")))
6cbb76823b7b planemo upload commit 3e8aad84b7fe9218370364b1aef3ea5779475f9f-dirty
proteore
parents: 23
diff changeset
582
6cbb76823b7b planemo upload commit 3e8aad84b7fe9218370364b1aef3ea5779475f9f-dirty
proteore
parents: 23
diff changeset
583 data_table_entry = dict(id=id, release=release_id, name = name, value = path)
6
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
584 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_nextprot_ref")
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
585
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
586 #######################################################################################################
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
587 # Main function
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
588 #######################################################################################################
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
589 def main():
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
590 parser = argparse.ArgumentParser()
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
591 parser.add_argument("--hpa", metavar = ("HPA_OPTION"))
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
592 parser.add_argument("--peptideatlas", metavar=("SAMPLE_CATEGORY_ID"))
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
593 parser.add_argument("--id_mapping", metavar = ("ID_MAPPING_SPECIES"))
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
594 parser.add_argument("--interactome", metavar = ("PPI"))
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
595 parser.add_argument("--species")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
596 parser.add_argument("--date")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
597 parser.add_argument("-o", "--output")
6
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
598 parser.add_argument("--database")
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
599 args = parser.parse_args()
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
600
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
601 data_manager_dict = {}
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
602 # Extract json file params
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
603 filename = args.output
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
604 params = from_json_string(open(filename).read())
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
605 target_directory = params[ 'output_data' ][0]['extra_files_path']
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
606 os.mkdir(target_directory)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
607
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
608 ## Download source files from HPA
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
609 try:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
610 hpa = args.hpa
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
611 except NameError:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
612 hpa = None
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
613 if hpa is not None:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
614 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
615 hpa = hpa.split(",")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
616 for hpa_tissue in hpa:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
617 HPA_sources(data_manager_dict, hpa_tissue, target_directory)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
618
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
619 ## Download source file from Peptide Atlas query
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
620 try:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
621 peptide_atlas = args.peptideatlas
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
622 date = args.date
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
623 except NameError:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
624 peptide_atlas = None
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
625 if peptide_atlas is not None:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
626 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
627 peptide_atlas = peptide_atlas.split(",")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
628 for pa_tissue in peptide_atlas:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
629 peptide_atlas_sources(data_manager_dict, pa_tissue, date, target_directory)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
630
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
631 ## Download ID_mapping source file from Uniprot
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
632 try:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
633 id_mapping=args.id_mapping
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
634 except NameError:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
635 id_mapping = None
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
636 if id_mapping is not None:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
637 id_mapping = id_mapping .split(",")
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
638 for species in id_mapping :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
639 id_mapping_sources(data_manager_dict, species, target_directory)
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
640
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
641 ## Download PPI ref files from biogrid/bioplex/humap
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
642 try:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
643 interactome=args.interactome
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
644 if interactome == "biogrid" :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
645 species=args.species
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
646 else :
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
647 species="Human"
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
648 except NameError:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
649 interactome=None
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
650 species=None
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
651 if interactome is not None and species is not None:
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
652 PPI_ref_files(data_manager_dict, species, interactome, target_directory)
6
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
653
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
654 ## Build nextprot ref file for add protein features
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
655 try:
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
656 database=args.database
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
657 except NameError:
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
658 database=None
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
659 if database is not None :
8
d5badf9de1b0 planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents: 6
diff changeset
660 Build_nextprot_ref_file(data_manager_dict,target_directory)
6
f281a1eb83d6 planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents: 5
diff changeset
661
0
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
662 #save info to json file
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
663 filename = args.output
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
664 open(filename, 'wb').write(to_json_string(data_manager_dict))
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
665
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
666 if __name__ == "__main__":
0a26460d7366 planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff changeset
667 main()