Mercurial > repos > proteore > proteore_data_manager
annotate data_manager/resource_building.py @ 5:429e7481c392 draft
planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
| author | proteore |
|---|---|
| date | Fri, 08 Mar 2019 04:09:43 -0500 |
| parents | ce01295cd601 |
| children | f281a1eb83d6 |
| 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") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
56 tissue_id = tissue_name.replace(" ","_").replace("/","-") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
57 |
|
1
0915249b8c4b
planemo upload commit 4dd1a2f7d196a1d2e70fab379a2c08367da0fe94-dirty
proteore
parents:
0
diff
changeset
|
58 |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
59 data_table_entry = dict(id=tissue_id, name = tissue_name, tissue = tissue, value = path) |
|
1
0915249b8c4b
planemo upload commit 4dd1a2f7d196a1d2e70fab379a2c08367da0fe94-dirty
proteore
parents:
0
diff
changeset
|
60 _add_data_table_entry(data_manager_dict, data_table_entry, table) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
61 |
|
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 # 2. Peptide Atlas |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
65 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
66 def peptide_atlas_sources(data_manager_dict, tissue, date, target_directory): |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
67 # 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
|
68 organism_id = "2" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
69 # Extract sample_category_id and output filename |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
70 tissue=tissue.split(".") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
71 sample_category_id = tissue[0] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
72 tissue_name = tissue[1] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
73 output_file = tissue_name+"_"+date + ".tsv" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
74 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
75 query="https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id="+ \ |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
76 sample_category_id+"&display_options=ShowAbundances&organism_id="+organism_id+ \ |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
77 "&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf\ |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
78 &QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
79 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
80 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
81 download = s.get(query) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
82 decoded_content = download.content.decode('utf-8') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
83 cr = csv.reader(decoded_content.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
84 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
85 uni_dict = build_dictionary(cr) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
86 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
87 #columns of data table peptide_atlas |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
88 tissue_id = tissue_name+"_"+date |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
89 name = tissue_id.replace("-","/").replace("_"," ") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
90 path = os.path.join(target_directory,output_file) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
91 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
92 with open(path,"w") as out : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
93 w = csv.writer(out,delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
94 w.writerow(["Uniprot_AC","nb_obs"]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
95 w.writerows(uni_dict.items()) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
96 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
97 data_table_entry = dict(id=tissue_id, name=name, value = path, tissue = tissue_name) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
98 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_peptide_atlas") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
99 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
100 #function to count the number of observations by uniprot id |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
101 def build_dictionary (csv) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
102 uni_dict = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
103 for line in csv : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
104 if "-" not in line[0] and check_uniprot_access(line[0]) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
105 if line[0] in uni_dict : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
106 uni_dict[line[0]] += int(line[5]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
107 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
108 uni_dict[line[0]] = int(line[5]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
109 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
110 return uni_dict |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
111 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
112 #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
|
113 def check_uniprot_access (id) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
114 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
|
115 if uniprot_pattern.match(id) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
116 return True |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
117 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
118 return False |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
119 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
120 def check_entrez_geneid (id) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
121 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
|
122 if entrez_pattern.match(id) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
123 return True |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
124 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
125 return False |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
126 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
127 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
128 # 3. ID mapping file |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
129 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
130 import ftplib, gzip |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
131 csv.field_size_limit(sys.maxsize) # to handle big files |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
132 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
133 def id_mapping_sources (data_manager_dict, species, target_directory) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
134 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
135 human = species == "Human" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
136 species_dict = { "Human" : "HUMAN_9606", "Mouse" : "MOUSE_10090", "Rat" : "RAT_10116" } |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
137 files=["idmapping_selected.tab.gz","idmapping.dat.gz"] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
138 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
139 #header |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
140 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"]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
141 else : tab = [["UniProt-AC","UniProt-ID","GeneID","RefSeq","GI","PDB","GO","PIR","MIM","UniGene","Ensembl_Gene","Ensembl_Transcript","Ensembl_Protein","BioGrid","STRING","KEGG"]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
142 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
143 #print("header ok") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
144 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
145 #get selected.tab and keep only ids of interest |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
146 selected_tab_file=species_dict[species]+"_"+files[0] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
147 tab_path = download_from_uniprot_ftp(selected_tab_file,target_directory) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
148 with gzip.open(tab_path,"rt") as select : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
149 tab_reader = csv.reader(select,delimiter="\t") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
150 for line in tab_reader : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
151 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
|
152 os.remove(tab_path) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
153 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
154 #print("selected_tab ok") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
155 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
156 """ |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
157 Supplementary ID to get from HUMAN_9606_idmapping.dat : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
158 -NextProt,BioGrid,STRING,KEGG |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
159 """ |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
160 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
161 #there's more id type for human |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
162 if human : ids = ['neXtProt','BioGrid','STRING','KEGG' ] #ids to get from dat_file |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
163 else : ids = ['BioGrid','STRING','KEGG' ] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
164 unidict = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
165 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
166 #keep only ids of interest in dictionaries |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
167 dat_file=species_dict[species]+"_"+files[1] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
168 dat_path = download_from_uniprot_ftp(dat_file,target_directory) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
169 with gzip.open(dat_path,"rt") as dat : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
170 dat_reader = csv.reader(dat,delimiter="\t") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
171 for line in dat_reader : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
172 uniprotID=line[0] #UniProtID as key |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
173 id_type=line[1] #ID type of corresponding id, key of sub-dictionnary |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
174 cor_id=line[2] #corresponding id |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
175 if "-" not in id_type : #we don't keep isoform |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
176 if id_type in ids and uniprotID in unidict : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
177 if id_type in unidict[uniprotID] : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
178 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
|
179 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
180 unidict[uniprotID].update({ id_type : cor_id }) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
181 elif id_type in ids : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
182 unidict[uniprotID]={id_type : cor_id} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
183 os.remove(dat_path) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
184 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
185 #print("dat_file ok") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
186 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
187 #add ids from idmapping.dat to the final tab |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
188 for line in tab[1:] : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
189 uniprotID=line[0] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
190 if human : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
191 if uniprotID in unidict : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
192 nextprot = access_dictionary(unidict,uniprotID,'neXtProt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
193 if nextprot != '' : nextprot = clean_nextprot_id(nextprot,line[0]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
194 line.extend([nextprot,access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'), |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
195 access_dictionary(unidict,uniprotID,'KEGG')]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
196 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
197 line.extend(["","","",""]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
198 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
199 if uniprotID in unidict : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
200 line.extend([access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'), |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
201 access_dictionary(unidict,uniprotID,'KEGG')]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
202 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
203 line.extend(["","",""]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
204 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
205 #print ("tab ok") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
206 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
207 #add missing nextprot ID for human |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
208 if human : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
209 #build next_dict |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
210 nextprot_ids = id_list_from_nextprot_ftp("nextprot_ac_list_all.txt",target_directory) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
211 next_dict = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
212 for nextid in nextprot_ids : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
213 next_dict[nextid.replace("NX_","")] = nextid |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
214 os.remove(os.path.join(target_directory,"nextprot_ac_list_all.txt")) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
215 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
216 #add missing nextprot ID |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
217 for line in tab[1:] : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
218 uniprotID=line[0] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
219 nextprotID=line[13] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
220 if nextprotID == '' and uniprotID in next_dict : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
221 line[13]=next_dict[uniprotID] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
222 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
223 output_file = species+"_id_mapping_"+ time.strftime("%d-%m-%Y") + ".tsv" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
224 path = os.path.join(target_directory,output_file) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
225 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
226 with open(path,"w") as out : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
227 w = csv.writer(out,delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
228 w.writerows(tab) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
229 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
230 name_dict={"Human" : "Homo sapiens", "Mouse" : "Mus musculus", "Rat" : "Rattus norvegicus"} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
231 name = species +" (" + name_dict[species]+" "+time.strftime("%d/%m/%Y")+")" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
232 id = species+"_id_mapping_"+ time.strftime("%d-%m-%Y") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
233 |
|
3
18d3b1ea7a18
planemo upload commit c6e1fd1f68e81ce9eea6ad66adee21070f2893ef-dirty
proteore
parents:
2
diff
changeset
|
234 data_table_entry = dict(id=id, name = name, species = species, value = path) |
|
2
2e34ee6d2d37
planemo upload commit 4dd1a2f7d196a1d2e70fab379a2c08367da0fe94-dirty
proteore
parents:
1
diff
changeset
|
235 _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
|
236 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
237 def download_from_uniprot_ftp(file,target_directory) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
238 ftp_dir = "pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
239 path = os.path.join(target_directory, file) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
240 ftp = ftplib.FTP("ftp.uniprot.org") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
241 ftp.login("anonymous", "anonymous") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
242 ftp.cwd(ftp_dir) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
243 ftp.retrbinary("RETR " + file, open(path, 'wb').write) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
244 ftp.quit() |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
245 return (path) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
246 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
247 def id_list_from_nextprot_ftp(file,target_directory) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
248 ftp_dir = "pub/current_release/ac_lists/" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
249 path = os.path.join(target_directory, file) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
250 ftp = ftplib.FTP("ftp.nextprot.org") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
251 ftp.login("anonymous", "anonymous") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
252 ftp.cwd(ftp_dir) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
253 ftp.retrbinary("RETR " + file, open(path, 'wb').write) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
254 ftp.quit() |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
255 with open(path,'r') as nextprot_ids : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
256 nextprot_ids = nextprot_ids.read().splitlines() |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
257 return (nextprot_ids) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
258 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
259 #return '' if there's no value in a dictionary, avoid error |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
260 def access_dictionary (dico,key1,key2) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
261 if key1 in dico : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
262 if key2 in dico[key1] : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
263 return (dico[key1][key2]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
264 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
265 return ("") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
266 #print (key2,"not in ",dico,"[",key1,"]") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
267 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
268 return ('') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
269 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
270 #if there are several nextprot ID for one uniprotID, return the uniprot like ID |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
271 def clean_nextprot_id (next_id,uniprotAc) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
272 if len(next_id.split(";")) > 1 : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
273 tmp = next_id.split(";") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
274 if "NX_"+uniprotAc in tmp : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
275 return ("NX_"+uniprotAc) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
276 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
277 return (tmp[1]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
278 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
279 return (next_id) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
280 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
281 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
282 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
283 # 4. Build protein interaction maps files |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
284 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
285 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
286 def get_interactant_name(line,dico): |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
287 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
288 if line[0] in dico : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
289 interactant_A = dico[line[0]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
290 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
291 interactant_A = "NA" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
292 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
293 if line[1] in dico : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
294 interactant_B = dico[line[1]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
295 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
296 interactant_B = "NA" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
297 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
298 return interactant_A, interactant_B |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
299 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
300 def PPI_ref_files(data_manager_dict, species, interactome, target_directory): |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
301 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
302 species_dict={'Human':'Homo sapiens',"Mouse":"Mus musculus","Rat":"Rattus norvegicus"} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
303 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
304 ##BioGRID |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
305 if interactome=="biogrid": |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
306 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
307 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
|
308 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
309 #download zip file |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
310 r = requests.get(tab2_link) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
311 with open("BioGRID.zip", "wb") as code: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
312 code.write(r.content) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
313 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
314 #unzip files |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
315 with zipfile.ZipFile("BioGRID.zip", 'r') as zip_ref: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
316 if not os.path.exists("tmp_BioGRID"): os.makedirs("tmp_BioGRID") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
317 zip_ref.extractall("tmp_BioGRID") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
318 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
319 #import file of interest and build dictionary |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
320 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
|
321 with open(file_path,"r") as handle : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
322 tab_file = csv.reader(handle,delimiter="\t") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
323 dico_network = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
324 GeneID_index=1 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
325 network_cols=[1,2,7,8,11,12,14,18,20] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
326 for line in tab_file : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
327 if line[GeneID_index] not in dico_network: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
328 dico_network[line[GeneID_index]]=[[line[i] for i in network_cols]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
329 else: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
330 dico_network[line[GeneID_index]].append([line[i] for i in network_cols]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
331 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
332 #delete tmp_BioGRID directory |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
333 os.remove("BioGRID.zip") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
334 shutil.rmtree("tmp_BioGRID", ignore_errors=True) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
335 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
336 #download NCBI2Reactome.txt file and build dictionary |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
337 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
338 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
339 r.encoding ="utf-8" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
340 tab_file = csv.reader(r.content.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
341 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
342 dico_nodes = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
343 geneid_index=0 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
344 pathway_description_index=3 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
345 species_index=5 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
346 for line in tab_file : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
347 if line[species_index]==species_dict[species]: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
348 if line[geneid_index] in dico_nodes : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
349 dico_nodes[line[geneid_index]].append(line[pathway_description_index]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
350 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
351 dico_nodes[line[geneid_index]] = [line[pathway_description_index]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
352 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
353 dico={} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
354 dico['network']=dico_network |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
355 dico['nodes']=dico_nodes |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
356 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
357 ##Bioplex |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
358 elif interactome=="bioplex": |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
359 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
360 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
361 r = s.get('http://bioplex.hms.harvard.edu/data/BioPlex_interactionList_v4a.tsv') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
362 r = r.content.decode('utf-8') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
363 bioplex = csv.reader(r.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
364 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
365 dico_network = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
366 dico_network["GeneID"]={} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
367 network_geneid_cols=[0,1,4,5,8] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
368 dico_network["UniProt-AC"]={} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
369 network_uniprot_cols=[2,3,4,5,8] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
370 dico_GeneID_to_UniProt = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
371 for line in bioplex : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
372 if line[0] not in dico_network["GeneID"]: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
373 dico_network["GeneID"][line[0]]=[[line[i] for i in network_geneid_cols]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
374 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
375 dico_network["GeneID"][line[0]].append([line[i] for i in network_geneid_cols]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
376 if line[1] not in dico_network["UniProt-AC"]: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
377 dico_network["UniProt-AC"][line[2]]=[[line[i] for i in network_uniprot_cols]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
378 else: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
379 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
|
380 dico_GeneID_to_UniProt[line[0]]=line[2] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
381 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
382 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
383 r = s.get('https://reactome.org/download/current/UniProt2Reactome.txt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
384 r.encoding ="utf-8" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
385 tab_file = csv.reader(r.content.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
386 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
387 dico_nodes_uniprot = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
388 uniProt_index=0 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
389 pathway_description_index=3 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
390 species_index=5 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
391 for line in tab_file : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
392 if line[species_index]==species_dict[species]: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
393 if line[uniProt_index] in dico_nodes_uniprot : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
394 dico_nodes_uniprot[line[uniProt_index]].append(line[pathway_description_index]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
395 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
396 dico_nodes_uniprot[line[uniProt_index]] = [line[pathway_description_index]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
397 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
398 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
399 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
400 r.encoding ="utf-8" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
401 tab_file = csv.reader(r.content.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
402 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
403 dico_nodes_geneid = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
404 geneid_index=0 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
405 pathway_description_index=3 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
406 species_index=5 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
407 for line in tab_file : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
408 if line[species_index]==species_dict[species]: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
409 if line[geneid_index] in dico_nodes_geneid : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
410 dico_nodes_geneid[line[geneid_index]].append(line[pathway_description_index]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
411 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
412 dico_nodes_geneid[line[geneid_index]] = [line[pathway_description_index]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
413 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
414 dico={} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
415 dico_nodes={} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
416 dico_nodes['GeneID']=dico_nodes_geneid |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
417 dico_nodes['UniProt-AC']=dico_nodes_uniprot |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
418 dico['network']=dico_network |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
419 dico['nodes']=dico_nodes |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
420 dico['convert']=dico_GeneID_to_UniProt |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
421 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
422 ##Humap |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
423 elif interactome=="humap": |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
424 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
425 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
426 r = s.get('http://proteincomplexes.org/static/downloads/nodeTable.txt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
427 r = r.content.decode('utf-8') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
428 humap_nodes = csv.reader(r.splitlines(), delimiter=',') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
429 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
430 dico_geneid_to_gene_name={} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
431 for line in humap_nodes : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
432 if check_entrez_geneid(line[4]): |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
433 if line[4] not in dico_geneid_to_gene_name: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
434 dico_geneid_to_gene_name[line[4]]=line[3] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
435 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
436 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
437 r = s.get('http://proteincomplexes.org/static/downloads/pairsWprob.txt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
438 r = r.content.decode('utf-8') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
439 humap = csv.reader(r.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
440 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
441 dico_network = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
442 for line in humap : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
443 if check_entrez_geneid(line[0]) and check_entrez_geneid(line[1]): |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
444 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
445 interactant_A, interactant_B = get_interactant_name(line,dico_geneid_to_gene_name) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
446 |
|
4
ce01295cd601
planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents:
3
diff
changeset
|
447 #first interactant (first column) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
448 if line[0] not in dico_network: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
449 dico_network[line[0]]=[line[:2]+[interactant_A,interactant_B,line[2]]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
450 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
451 dico_network[line[0]].append(line[:2]+[interactant_A,interactant_B,line[2]]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
452 |
|
4
ce01295cd601
planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents:
3
diff
changeset
|
453 #second interactant (second column) |
|
ce01295cd601
planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents:
3
diff
changeset
|
454 if line[1] not in dico_network: |
|
5
429e7481c392
planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents:
4
diff
changeset
|
455 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
|
456 else : |
|
5
429e7481c392
planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents:
4
diff
changeset
|
457 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
|
458 |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
459 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
460 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
461 r.encoding ="utf-8" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
462 tab_file = csv.reader(r.content.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
463 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
464 dico_nodes = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
465 geneid_index=0 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
466 pathway_description_index=3 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
467 species_index=5 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
468 for line in tab_file : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
469 if line[species_index]==species_dict[species]: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
470 #Fill dictionary with pathways |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
471 if line[geneid_index] in dico_nodes : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
472 dico_nodes[line[geneid_index]].append(line[pathway_description_index]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
473 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
474 dico_nodes[line[geneid_index]] = [line[pathway_description_index]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
475 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
476 dico={} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
477 dico['network']=dico_network |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
478 dico['nodes']=dico_nodes |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
479 dico['gene_name']=dico_geneid_to_gene_name |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
480 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
481 #writing output |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
482 output_file = species+'_'+interactome+'_'+ time.strftime("%d-%m-%Y") + ".json" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
483 path = os.path.join(target_directory,output_file) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
484 name = species+" ("+species_dict[species]+") "+time.strftime("%d/%m/%Y") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
485 id = species+"_"+interactome+"_"+ time.strftime("%d-%m-%Y") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
486 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
487 with open(path, 'w') as handle: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
488 json.dump(dico, handle, sort_keys=True) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
489 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
490 data_table_entry = dict(id=id, name = name, species = species, value = path) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
491 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_"+interactome+"_dictionaries") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
492 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
493 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
494 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
495 # Main function |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
496 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
497 def main(): |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
498 parser = argparse.ArgumentParser() |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
499 parser.add_argument("--hpa", metavar = ("HPA_OPTION")) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
500 parser.add_argument("--peptideatlas", metavar=("SAMPLE_CATEGORY_ID")) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
501 parser.add_argument("--id_mapping", metavar = ("ID_MAPPING_SPECIES")) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
502 parser.add_argument("--interactome", metavar = ("PPI")) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
503 parser.add_argument("--species") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
504 parser.add_argument("--date") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
505 parser.add_argument("-o", "--output") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
506 args = parser.parse_args() |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
507 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
508 data_manager_dict = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
509 # Extract json file params |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
510 filename = args.output |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
511 params = from_json_string(open(filename).read()) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
512 target_directory = params[ 'output_data' ][0]['extra_files_path'] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
513 os.mkdir(target_directory) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
514 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
515 ## Download source files from HPA |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
516 try: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
517 hpa = args.hpa |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
518 except NameError: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
519 hpa = None |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
520 if hpa is not None: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
521 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
522 hpa = hpa.split(",") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
523 for hpa_tissue in hpa: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
524 HPA_sources(data_manager_dict, hpa_tissue, target_directory) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
525 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
526 ## Download source file from Peptide Atlas query |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
527 try: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
528 peptide_atlas = args.peptideatlas |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
529 date = args.date |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
530 except NameError: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
531 peptide_atlas = None |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
532 if peptide_atlas is not None: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
533 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
534 peptide_atlas = peptide_atlas.split(",") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
535 for pa_tissue in peptide_atlas: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
536 peptide_atlas_sources(data_manager_dict, pa_tissue, date, target_directory) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
537 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
538 ## Download ID_mapping source file from Uniprot |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
539 try: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
540 id_mapping=args.id_mapping |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
541 except NameError: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
542 id_mapping = None |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
543 if id_mapping is not None: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
544 id_mapping = id_mapping .split(",") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
545 for species in id_mapping : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
546 id_mapping_sources(data_manager_dict, species, target_directory) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
547 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
548 ## Download PPI ref files from biogrid/bioplex/humap |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
549 try: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
550 interactome=args.interactome |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
551 if interactome == "biogrid" : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
552 species=args.species |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
553 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
554 species="Human" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
555 except NameError: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
556 interactome=None |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
557 species=None |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
558 if interactome is not None and species is not None: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
559 PPI_ref_files(data_manager_dict, species, interactome, target_directory) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
560 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
561 #save info to json file |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
562 filename = args.output |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
563 open(filename, 'wb').write(to_json_string(data_manager_dict)) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
564 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
565 if __name__ == "__main__": |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
566 main() |
