Mercurial > repos > proteore > proteore_data_manager
annotate data_manager/resource_building.py @ 48:09d9fd9b2d8c draft
"planemo upload commit 714703437f238e7295980f96620365b801fd435c-dirty"
| author | proteore |
|---|---|
| date | Fri, 31 Jan 2020 15:51:40 +0000 |
| parents | 488e015576bc |
| children | 1c40456be1db |
| 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 |
|
42
f6a6a70712c4
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
41
diff
changeset
|
6 import os, shutil, sys, argparse, requests, time, csv, re, json, shutil, zipfile, subprocess |
|
0
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" |
|
48
09d9fd9b2d8c
"planemo upload commit 714703437f238e7295980f96620365b801fd435c-dirty"
proteore
parents:
45
diff
changeset
|
51 elif tissue == "HPA_RNA_tissue": |
|
09d9fd9b2d8c
"planemo upload commit 714703437f238e7295980f96620365b801fd435c-dirty"
proteore
parents:
45
diff
changeset
|
52 tissue_name = "HPA RNA tissue" |
|
09d9fd9b2d8c
"planemo upload commit 714703437f238e7295980f96620365b801fd435c-dirty"
proteore
parents:
45
diff
changeset
|
53 url = "https://www.proteinatlas.org/download/normal_tissue.tsv.zip" |
|
09d9fd9b2d8c
"planemo upload commit 714703437f238e7295980f96620365b801fd435c-dirty"
proteore
parents:
45
diff
changeset
|
54 table = "proteore_protein_atlas_rna_tissue" |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
55 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
56 output_file = tissue +"_"+ time.strftime("%d-%m-%Y") + ".tsv" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
57 path = os.path.join(target_directory, output_file) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
58 unzip(url, path) #download and save file |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
59 tissue_name = tissue_name + " " + time.strftime("%d/%m/%Y") |
|
26
bf6940ff60a8
planemo upload commit 0290724216a2c445b4e28842153b84a1b28f4e9a-dirty
proteore
parents:
24
diff
changeset
|
60 release = tissue_name.replace(" ","_").replace("/","-") |
|
bf6940ff60a8
planemo upload commit 0290724216a2c445b4e28842153b84a1b28f4e9a-dirty
proteore
parents:
24
diff
changeset
|
61 id = str(10000000000 - int(time.strftime("%Y%m%d"))) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
62 |
|
1
0915249b8c4b
planemo upload commit 4dd1a2f7d196a1d2e70fab379a2c08367da0fe94-dirty
proteore
parents:
0
diff
changeset
|
63 |
|
26
bf6940ff60a8
planemo upload commit 0290724216a2c445b4e28842153b84a1b28f4e9a-dirty
proteore
parents:
24
diff
changeset
|
64 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
|
65 _add_data_table_entry(data_manager_dict, data_table_entry, table) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
66 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
67 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
68 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
69 # 2. Peptide Atlas |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
70 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
71 def peptide_atlas_sources(data_manager_dict, tissue, date, target_directory): |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
72 # 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
|
73 organism_id = "2" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
74 # Extract sample_category_id and output filename |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
75 tissue=tissue.split(".") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
76 sample_category_id = tissue[0] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
77 tissue_name = tissue[1] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
78 output_file = tissue_name+"_"+date + ".tsv" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
79 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
80 query="https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id="+ \ |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
81 sample_category_id+"&display_options=ShowAbundances&organism_id="+organism_id+ \ |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
82 "&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf\ |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
83 &QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
84 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
85 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
86 download = s.get(query) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
87 decoded_content = download.content.decode('utf-8') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
88 cr = csv.reader(decoded_content.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
89 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
90 uni_dict = build_dictionary(cr) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
91 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
92 #columns of data table peptide_atlas |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
93 tissue_id = tissue_name+"_"+date |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
94 name = tissue_id.replace("-","/").replace("_"," ") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
95 path = os.path.join(target_directory,output_file) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
96 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
97 with open(path,"w") as out : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
98 w = csv.writer(out,delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
99 w.writerow(["Uniprot_AC","nb_obs"]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
100 w.writerows(uni_dict.items()) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
101 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
102 data_table_entry = dict(id=tissue_id, name=name, value = path, tissue = tissue_name) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
103 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_peptide_atlas") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
104 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
105 #function to count the number of observations by uniprot id |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
106 def build_dictionary (csv) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
107 uni_dict = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
108 for line in csv : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
109 if "-" not in line[0] and check_uniprot_access(line[0]) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
110 if line[0] in uni_dict : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
111 uni_dict[line[0]] += int(line[5]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
112 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
113 uni_dict[line[0]] = int(line[5]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
114 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
115 return uni_dict |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
116 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
117 #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
|
118 def check_uniprot_access (id) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
119 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
|
120 if uniprot_pattern.match(id) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
121 return True |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
122 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
123 return False |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
124 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
125 def check_entrez_geneid (id) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
126 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
|
127 if entrez_pattern.match(id) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
128 return True |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
129 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
130 return False |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
131 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
132 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
133 # 3. ID mapping file |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
134 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
135 import ftplib, gzip |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
136 csv.field_size_limit(sys.maxsize) # to handle big files |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
137 |
|
31
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
138 def id_mapping_sources (data_manager_dict, species, target_directory, tool_data_path) : |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
139 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
140 human = species == "Human" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
141 species_dict = { "Human" : "HUMAN_9606", "Mouse" : "MOUSE_10090", "Rat" : "RAT_10116" } |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
142 files=["idmapping_selected.tab.gz","idmapping.dat.gz"] |
|
36
b18cb0d11971
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
35
diff
changeset
|
143 archive = os.path.join(tool_data_path, "id_mapping/ID_mapping_archive_"+species+"_"+str(time.strftime("%Y%m%d"))) |
|
38
6daab0a711e2
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
37
diff
changeset
|
144 if os.path.isdir(archive) is False : os.mkdir(archive) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
145 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
146 #header |
|
30
a6cabd3ab71f
"planemo upload commit b89f1921a1759139b452c6fac1ad7ee01b6b633d-dirty"
proteore
parents:
29
diff
changeset
|
147 if human : tab = [["UniProt-AC","UniProt-AC_reviewed","UniProt-ID","GeneID","RefSeq","GI","PDB","GO","PIR","MIM","UniGene","Ensembl_Gene","Ensembl_Transcript","Ensembl_Protein","neXtProt","BioGrid","STRING","KEGG",'Gene_Name']] |
|
a6cabd3ab71f
"planemo upload commit b89f1921a1759139b452c6fac1ad7ee01b6b633d-dirty"
proteore
parents:
29
diff
changeset
|
148 else : tab = [["UniProt-AC","UniProt-AC_reviewed","UniProt-ID","GeneID","RefSeq","GI","PDB","GO","PIR","MIM","UniGene","Ensembl_Gene","Ensembl_Transcript","Ensembl_Protein","BioGrid","STRING","KEGG",'Gene_Name']] |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
149 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
150 #get selected.tab and keep only ids of interest |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
151 selected_tab_file=species_dict[species]+"_"+files[0] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
152 tab_path = download_from_uniprot_ftp(selected_tab_file,target_directory) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
153 with gzip.open(tab_path,"rt") as select : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
154 tab_reader = csv.reader(select,delimiter="\t") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
155 for line in tab_reader : |
|
30
a6cabd3ab71f
"planemo upload commit b89f1921a1759139b452c6fac1ad7ee01b6b633d-dirty"
proteore
parents:
29
diff
changeset
|
156 tab.append([line[0]]+[line[i] for i in [0,1,2,3,4,5,6,11,13,14,18,19,20]]) |
|
41
fe21769281fa
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
40
diff
changeset
|
157 if os.path.exists(os.path.join(archive,tab_path.split("/")[-1])) : os.remove(os.path.join(archive,tab_path.split("/")[-1])) |
|
31
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
158 shutil.move(tab_path, archive) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
159 #print("selected_tab ok") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
160 |
|
20
29cf75c83618
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
16
diff
changeset
|
161 #get uniprot-AC reviewed |
|
22
4296fb613d9d
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
21
diff
changeset
|
162 organism = species_dict[species].split("_")[1] |
|
4296fb613d9d
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
21
diff
changeset
|
163 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
|
164 |
|
29cf75c83618
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
16
diff
changeset
|
165 with requests.Session() as s: |
|
29cf75c83618
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
16
diff
changeset
|
166 download = s.get(query) |
|
29cf75c83618
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
16
diff
changeset
|
167 decoded_content = download.content.decode('utf-8') |
|
29cf75c83618
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
16
diff
changeset
|
168 uniprot_reviewed_list = decoded_content.splitlines() |
|
29cf75c83618
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
16
diff
changeset
|
169 |
|
31
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
170 #save reviewed list |
|
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
171 reviewed_list_path = os.path.join(archive,'uniprot_reviewed_list.txt') |
|
34
06b3181206b4
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
33
diff
changeset
|
172 with open(reviewed_list_path,'w') as reviewed_list_file: |
|
31
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
173 for id in uniprot_reviewed_list: |
|
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
174 reviewed_list_file.write(id+"\n") |
|
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
175 |
|
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
176 #remove unreviewed uniprot-AC |
|
20
29cf75c83618
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
16
diff
changeset
|
177 for line in tab[1:]: |
|
30
a6cabd3ab71f
"planemo upload commit b89f1921a1759139b452c6fac1ad7ee01b6b633d-dirty"
proteore
parents:
29
diff
changeset
|
178 UniProtAC = line[1] |
|
28
808c8493ed4f
planemo upload commit 5c2c274361c0daceae1f678eca0e6c0d5b4ba4f7-dirty
proteore
parents:
27
diff
changeset
|
179 if UniProtAC not in uniprot_reviewed_list : |
|
29
9a40b72414de
planemo upload commit 3c02acc07df862410ce979ce63e3d14f10edc50f-dirty
proteore
parents:
28
diff
changeset
|
180 line[1]="" |
|
20
29cf75c83618
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
16
diff
changeset
|
181 |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
182 """ |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
183 Supplementary ID to get from HUMAN_9606_idmapping.dat : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
184 -NextProt,BioGrid,STRING,KEGG |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
185 """ |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
186 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
187 #there's more id type for human |
|
30
a6cabd3ab71f
"planemo upload commit b89f1921a1759139b452c6fac1ad7ee01b6b633d-dirty"
proteore
parents:
29
diff
changeset
|
188 if human : ids = ['neXtProt','BioGrid','STRING','KEGG','Gene_Name' ] #ids to get from dat_file |
|
a6cabd3ab71f
"planemo upload commit b89f1921a1759139b452c6fac1ad7ee01b6b633d-dirty"
proteore
parents:
29
diff
changeset
|
189 else : ids = ['BioGrid','STRING','KEGG','Gene_Name' ] |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
190 unidict = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
191 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
192 #keep only ids of interest in dictionaries |
|
40
a7bcc9a0a456
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
39
diff
changeset
|
193 dat_file = species_dict[species]+"_"+files[1] |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
194 dat_path = download_from_uniprot_ftp(dat_file,target_directory) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
195 with gzip.open(dat_path,"rt") as dat : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
196 dat_reader = csv.reader(dat,delimiter="\t") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
197 for line in dat_reader : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
198 uniprotID=line[0] #UniProtID as key |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
199 id_type=line[1] #ID type of corresponding id, key of sub-dictionnary |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
200 cor_id=line[2] #corresponding id |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
201 if "-" not in id_type : #we don't keep isoform |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
202 if id_type in ids and uniprotID in unidict : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
203 if id_type in unidict[uniprotID] : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
204 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
|
205 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
206 unidict[uniprotID].update({ id_type : cor_id }) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
207 elif id_type in ids : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
208 unidict[uniprotID]={id_type : cor_id} |
|
41
fe21769281fa
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
40
diff
changeset
|
209 if os.path.exists(os.path.join(archive,dat_path.split("/")[-1])) : os.remove(os.path.join(archive,dat_path.split("/")[-1])) |
|
31
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
210 shutil.move(dat_path, archive) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
211 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
212 #print("dat_file ok") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
213 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
214 #add ids from idmapping.dat to the final tab |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
215 for line in tab[1:] : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
216 uniprotID=line[0] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
217 if human : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
218 if uniprotID in unidict : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
219 nextprot = access_dictionary(unidict,uniprotID,'neXtProt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
220 if nextprot != '' : nextprot = clean_nextprot_id(nextprot,line[0]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
221 line.extend([nextprot,access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'), |
|
30
a6cabd3ab71f
"planemo upload commit b89f1921a1759139b452c6fac1ad7ee01b6b633d-dirty"
proteore
parents:
29
diff
changeset
|
222 access_dictionary(unidict,uniprotID,'KEGG'),access_dictionary(unidict,uniprotID,'Gene_Name')]) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
223 else : |
|
30
a6cabd3ab71f
"planemo upload commit b89f1921a1759139b452c6fac1ad7ee01b6b633d-dirty"
proteore
parents:
29
diff
changeset
|
224 line.extend(["","","","",""]) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
225 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
226 if uniprotID in unidict : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
227 line.extend([access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'), |
|
30
a6cabd3ab71f
"planemo upload commit b89f1921a1759139b452c6fac1ad7ee01b6b633d-dirty"
proteore
parents:
29
diff
changeset
|
228 access_dictionary(unidict,uniprotID,'KEGG'),access_dictionary(unidict,uniprotID,'Gene_Name')]) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
229 else : |
|
30
a6cabd3ab71f
"planemo upload commit b89f1921a1759139b452c6fac1ad7ee01b6b633d-dirty"
proteore
parents:
29
diff
changeset
|
230 line.extend(["","","",""]) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
231 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
232 #print ("tab ok") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
233 |
|
20
29cf75c83618
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
16
diff
changeset
|
234 #add missing nextprot ID for human or replace old ones |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
235 if human : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
236 #build next_dict |
|
31
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
237 nextprot_path = id_list_from_nextprot_ftp("nextprot_ac_list_all.txt",target_directory) |
|
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
238 with open(nextprot_path,'r') as nextprot_ids : |
|
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
239 nextprot_ids = nextprot_ids.read().splitlines() |
|
41
fe21769281fa
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
40
diff
changeset
|
240 if os.path.exists(os.path.join(archive,nextprot_path.split("/")[-1])) : os.remove(os.path.join(archive,nextprot_path.split("/")[-1])) |
|
31
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
241 shutil.move(nextprot_path,archive) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
242 next_dict = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
243 for nextid in nextprot_ids : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
244 next_dict[nextid.replace("NX_","")] = nextid |
|
45
488e015576bc
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
44
diff
changeset
|
245 #os.remove(os.path.join(target_directory,"nextprot_ac_list_all.txt")) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
246 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
247 #add missing nextprot ID |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
248 for line in tab[1:] : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
249 uniprotID=line[0] |
|
30
a6cabd3ab71f
"planemo upload commit b89f1921a1759139b452c6fac1ad7ee01b6b633d-dirty"
proteore
parents:
29
diff
changeset
|
250 nextprotID=line[14] |
|
20
29cf75c83618
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
16
diff
changeset
|
251 if uniprotID in next_dict and (nextprotID == '' or (nextprotID != "NX_"+uniprotID and next_dict[uniprotID] == "NX_"+uniprotID)) : |
|
30
a6cabd3ab71f
"planemo upload commit b89f1921a1759139b452c6fac1ad7ee01b6b633d-dirty"
proteore
parents:
29
diff
changeset
|
252 line[14]=next_dict[uniprotID] |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
253 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
254 output_file = species+"_id_mapping_"+ time.strftime("%d-%m-%Y") + ".tsv" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
255 path = os.path.join(target_directory,output_file) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
256 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
257 with open(path,"w") as out : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
258 w = csv.writer(out,delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
259 w.writerows(tab) |
|
31
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
260 |
|
36
b18cb0d11971
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
35
diff
changeset
|
261 subprocess.call(['tar', '-czvf', archive+".tar.gz", archive]) |
|
44
7ebbf851598e
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
43
diff
changeset
|
262 shutil.rmtree(archive, ignore_errors=True) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
263 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
264 name_dict={"Human" : "Homo sapiens", "Mouse" : "Mus musculus", "Rat" : "Rattus norvegicus"} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
265 name = species +" (" + name_dict[species]+" "+time.strftime("%d/%m/%Y")+")" |
|
21
026177e4ff4b
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
20
diff
changeset
|
266 release = species+"_id_mapping_"+ time.strftime("%d-%m-%Y") |
|
23
ebdd4961c6c2
planemo upload commit 03015f58ac7e7ba3cc44ba0d8899eacfaaf5a134-dirty
proteore
parents:
22
diff
changeset
|
267 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
|
268 |
|
21
026177e4ff4b
planemo upload commit e05ccbf13c33b97ab441e2f8bc4b5bc746a378df-dirty
proteore
parents:
20
diff
changeset
|
269 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
|
270 _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
|
271 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
272 def download_from_uniprot_ftp(file,target_directory) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
273 ftp_dir = "pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
274 path = os.path.join(target_directory, file) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
275 ftp = ftplib.FTP("ftp.uniprot.org") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
276 ftp.login("anonymous", "anonymous") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
277 ftp.cwd(ftp_dir) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
278 ftp.retrbinary("RETR " + file, open(path, 'wb').write) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
279 ftp.quit() |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
280 return (path) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
281 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
282 def id_list_from_nextprot_ftp(file,target_directory) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
283 ftp_dir = "pub/current_release/ac_lists/" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
284 path = os.path.join(target_directory, file) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
285 ftp = ftplib.FTP("ftp.nextprot.org") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
286 ftp.login("anonymous", "anonymous") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
287 ftp.cwd(ftp_dir) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
288 ftp.retrbinary("RETR " + file, open(path, 'wb').write) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
289 ftp.quit() |
|
31
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
290 |
|
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
291 return (path) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
292 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
293 #return '' if there's no value in a dictionary, avoid error |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
294 def access_dictionary (dico,key1,key2) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
295 if key1 in dico : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
296 if key2 in dico[key1] : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
297 return (dico[key1][key2]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
298 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
299 return ("") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
300 #print (key2,"not in ",dico,"[",key1,"]") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
301 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
302 return ('') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
303 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
304 #if there are several nextprot ID for one uniprotID, return the uniprot like ID |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
305 def clean_nextprot_id (next_id,uniprotAc) : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
306 if len(next_id.split(";")) > 1 : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
307 tmp = next_id.split(";") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
308 if "NX_"+uniprotAc in tmp : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
309 return ("NX_"+uniprotAc) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
310 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
311 return (tmp[1]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
312 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
313 return (next_id) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
314 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
315 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
316 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
317 # 4. Build protein interaction maps files |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
318 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
319 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
320 def get_interactant_name(line,dico): |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
321 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
322 if line[0] in dico : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
323 interactant_A = dico[line[0]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
324 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
325 interactant_A = "NA" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
326 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
327 if line[1] in dico : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
328 interactant_B = dico[line[1]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
329 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
330 interactant_B = "NA" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
331 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
332 return interactant_A, interactant_B |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
333 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
334 def PPI_ref_files(data_manager_dict, species, interactome, target_directory): |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
335 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
336 species_dict={'Human':'Homo sapiens',"Mouse":"Mus musculus","Rat":"Rattus norvegicus"} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
337 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
338 ##BioGRID |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
339 if interactome=="biogrid": |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
340 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
341 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
|
342 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
343 #download zip file |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
344 r = requests.get(tab2_link) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
345 with open("BioGRID.zip", "wb") as code: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
346 code.write(r.content) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
347 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
348 #unzip files |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
349 with zipfile.ZipFile("BioGRID.zip", 'r') as zip_ref: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
350 if not os.path.exists("tmp_BioGRID"): os.makedirs("tmp_BioGRID") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
351 zip_ref.extractall("tmp_BioGRID") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
352 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
353 #import file of interest and build dictionary |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
354 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
|
355 with open(file_path,"r") as handle : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
356 tab_file = csv.reader(handle,delimiter="\t") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
357 dico_network = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
358 GeneID_index=1 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
359 network_cols=[1,2,7,8,11,12,14,18,20] |
|
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[GeneID_index] not in dico_network: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
362 dico_network[line[GeneID_index]]=[[line[i] for i in network_cols]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
363 else: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
364 dico_network[line[GeneID_index]].append([line[i] for i in network_cols]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
365 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
366 #delete tmp_BioGRID directory |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
367 os.remove("BioGRID.zip") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
368 shutil.rmtree("tmp_BioGRID", ignore_errors=True) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
369 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
370 #download NCBI2Reactome.txt file and build dictionary |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
371 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
372 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
373 r.encoding ="utf-8" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
374 tab_file = csv.reader(r.content.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
375 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
376 dico_nodes = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
377 geneid_index=0 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
378 pathway_description_index=3 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
379 species_index=5 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
380 for line in tab_file : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
381 if line[species_index]==species_dict[species]: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
382 if line[geneid_index] in dico_nodes : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
383 dico_nodes[line[geneid_index]].append(line[pathway_description_index]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
384 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
385 dico_nodes[line[geneid_index]] = [line[pathway_description_index]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
386 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
387 dico={} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
388 dico['network']=dico_network |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
389 dico['nodes']=dico_nodes |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
390 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
391 ##Bioplex |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
392 elif interactome=="bioplex": |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
393 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
394 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
395 r = s.get('http://bioplex.hms.harvard.edu/data/BioPlex_interactionList_v4a.tsv') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
396 r = r.content.decode('utf-8') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
397 bioplex = csv.reader(r.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
398 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
399 dico_network = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
400 dico_network["GeneID"]={} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
401 network_geneid_cols=[0,1,4,5,8] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
402 dico_network["UniProt-AC"]={} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
403 network_uniprot_cols=[2,3,4,5,8] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
404 dico_GeneID_to_UniProt = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
405 for line in bioplex : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
406 if line[0] not in dico_network["GeneID"]: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
407 dico_network["GeneID"][line[0]]=[[line[i] for i in network_geneid_cols]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
408 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
409 dico_network["GeneID"][line[0]].append([line[i] for i in network_geneid_cols]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
410 if line[1] not in dico_network["UniProt-AC"]: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
411 dico_network["UniProt-AC"][line[2]]=[[line[i] for i in network_uniprot_cols]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
412 else: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
413 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
|
414 dico_GeneID_to_UniProt[line[0]]=line[2] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
415 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
416 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
417 r = s.get('https://reactome.org/download/current/UniProt2Reactome.txt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
418 r.encoding ="utf-8" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
419 tab_file = csv.reader(r.content.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
420 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
421 dico_nodes_uniprot = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
422 uniProt_index=0 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
423 pathway_description_index=3 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
424 species_index=5 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
425 for line in tab_file : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
426 if line[species_index]==species_dict[species]: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
427 if line[uniProt_index] in dico_nodes_uniprot : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
428 dico_nodes_uniprot[line[uniProt_index]].append(line[pathway_description_index]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
429 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
430 dico_nodes_uniprot[line[uniProt_index]] = [line[pathway_description_index]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
431 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
432 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
433 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
434 r.encoding ="utf-8" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
435 tab_file = csv.reader(r.content.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
436 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
437 dico_nodes_geneid = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
438 geneid_index=0 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
439 pathway_description_index=3 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
440 species_index=5 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
441 for line in tab_file : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
442 if line[species_index]==species_dict[species]: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
443 if line[geneid_index] in dico_nodes_geneid : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
444 dico_nodes_geneid[line[geneid_index]].append(line[pathway_description_index]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
445 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
446 dico_nodes_geneid[line[geneid_index]] = [line[pathway_description_index]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
447 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
448 dico={} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
449 dico_nodes={} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
450 dico_nodes['GeneID']=dico_nodes_geneid |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
451 dico_nodes['UniProt-AC']=dico_nodes_uniprot |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
452 dico['network']=dico_network |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
453 dico['nodes']=dico_nodes |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
454 dico['convert']=dico_GeneID_to_UniProt |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
455 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
456 ##Humap |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
457 elif interactome=="humap": |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
458 |
|
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('http://proteincomplexes.org/static/downloads/nodeTable.txt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
461 r = r.content.decode('utf-8') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
462 humap_nodes = csv.reader(r.splitlines(), delimiter=',') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
463 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
464 dico_geneid_to_gene_name={} |
|
9
cdd29444e0af
planemo upload commit 71363136045353f422ff98219c1eb84f6fc6193a-dirty
proteore
parents:
8
diff
changeset
|
465 dico_protein_name={} |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
466 for line in humap_nodes : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
467 if check_entrez_geneid(line[4]): |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
468 if line[4] not in dico_geneid_to_gene_name: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
469 dico_geneid_to_gene_name[line[4]]=line[3] |
|
9
cdd29444e0af
planemo upload commit 71363136045353f422ff98219c1eb84f6fc6193a-dirty
proteore
parents:
8
diff
changeset
|
470 if line[4] not in dico_protein_name: |
|
cdd29444e0af
planemo upload commit 71363136045353f422ff98219c1eb84f6fc6193a-dirty
proteore
parents:
8
diff
changeset
|
471 dico_protein_name[line[4]]=line[5] |
|
cdd29444e0af
planemo upload commit 71363136045353f422ff98219c1eb84f6fc6193a-dirty
proteore
parents:
8
diff
changeset
|
472 |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
473 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
474 r = s.get('http://proteincomplexes.org/static/downloads/pairsWprob.txt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
475 r = r.content.decode('utf-8') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
476 humap = csv.reader(r.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
477 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
478 dico_network = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
479 for line in humap : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
480 if check_entrez_geneid(line[0]) and check_entrez_geneid(line[1]): |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
481 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
482 interactant_A, interactant_B = get_interactant_name(line,dico_geneid_to_gene_name) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
483 |
|
4
ce01295cd601
planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents:
3
diff
changeset
|
484 #first interactant (first column) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
485 if line[0] not in dico_network: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
486 dico_network[line[0]]=[line[:2]+[interactant_A,interactant_B,line[2]]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
487 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
488 dico_network[line[0]].append(line[:2]+[interactant_A,interactant_B,line[2]]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
489 |
|
4
ce01295cd601
planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents:
3
diff
changeset
|
490 #second interactant (second column) |
|
ce01295cd601
planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents:
3
diff
changeset
|
491 if line[1] not in dico_network: |
|
5
429e7481c392
planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents:
4
diff
changeset
|
492 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
|
493 else : |
|
5
429e7481c392
planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents:
4
diff
changeset
|
494 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
|
495 |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
496 with requests.Session() as s: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
497 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
498 r.encoding ="utf-8" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
499 tab_file = csv.reader(r.content.splitlines(), delimiter='\t') |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
500 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
501 dico_nodes = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
502 geneid_index=0 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
503 pathway_description_index=3 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
504 species_index=5 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
505 for line in tab_file : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
506 if line[species_index]==species_dict[species]: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
507 #Fill dictionary with pathways |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
508 if line[geneid_index] in dico_nodes : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
509 dico_nodes[line[geneid_index]].append(line[pathway_description_index]) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
510 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
511 dico_nodes[line[geneid_index]] = [line[pathway_description_index]] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
512 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
513 dico={} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
514 dico['network']=dico_network |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
515 dico['nodes']=dico_nodes |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
516 dico['gene_name']=dico_geneid_to_gene_name |
|
9
cdd29444e0af
planemo upload commit 71363136045353f422ff98219c1eb84f6fc6193a-dirty
proteore
parents:
8
diff
changeset
|
517 dico['protein_name']=dico_protein_name |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
518 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
519 #writing output |
|
27
9a400ce8e4e6
planemo upload commit bb113d19d3a756f70784e6a1433902888686ed96-dirty
proteore
parents:
26
diff
changeset
|
520 output_file = species+'_'+interactome+'_'+ time.strftime("%Y-%m-%d") + ".json" |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
521 path = os.path.join(target_directory,output_file) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
522 name = species+" ("+species_dict[species]+") "+time.strftime("%d/%m/%Y") |
|
27
9a400ce8e4e6
planemo upload commit bb113d19d3a756f70784e6a1433902888686ed96-dirty
proteore
parents:
26
diff
changeset
|
523 release = species+"_"+interactome+"_"+ time.strftime("%Y-%m-%d") |
|
26
bf6940ff60a8
planemo upload commit 0290724216a2c445b4e28842153b84a1b28f4e9a-dirty
proteore
parents:
24
diff
changeset
|
524 id = str(10000000000 - int(time.strftime("%Y%m%d"))) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
525 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
526 with open(path, 'w') as handle: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
527 json.dump(dico, handle, sort_keys=True) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
528 |
|
26
bf6940ff60a8
planemo upload commit 0290724216a2c445b4e28842153b84a1b28f4e9a-dirty
proteore
parents:
24
diff
changeset
|
529 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
|
530 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_"+interactome+"_dictionaries") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
531 |
|
6
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
532 ####################################################################################################### |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
533 # 5. nextprot (add protein features) |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
534 ####################################################################################################### |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
535 |
|
8
d5badf9de1b0
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
6
diff
changeset
|
536 def Build_nextprot_ref_file(data_manager_dict,target_directory): |
|
6
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
537 nextprot_ids_file = "nextprot_ac_list_all.txt" |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
538 ids = id_list_from_nextprot_ftp(nextprot_ids_file,target_directory) |
|
12
f6afaa1f562c
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents:
11
diff
changeset
|
539 |
|
f6afaa1f562c
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents:
11
diff
changeset
|
540 output_file = 'nextprot_ref_'+ time.strftime("%d-%m-%Y") + ".tsv" |
|
f6afaa1f562c
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents:
11
diff
changeset
|
541 path = os.path.join(target_directory,output_file) |
|
f6afaa1f562c
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents:
11
diff
changeset
|
542 name = "neXtProt release "+time.strftime("%d-%m-%Y") |
|
16
f75c525e0a4a
planemo upload commit 04868e380c43447fac1309fc292785d67863a87b-dirty
proteore
parents:
15
diff
changeset
|
543 release_id = "nextprot_ref_"+time.strftime("%d-%m-%Y") |
|
12
f6afaa1f562c
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents:
11
diff
changeset
|
544 |
|
13
098693479a9d
planemo upload commit 64366857c1f29dd79a77dc331eb0153cb94cdb11
proteore
parents:
12
diff
changeset
|
545 output = open(path, 'w') |
|
098693479a9d
planemo upload commit 64366857c1f29dd79a77dc331eb0153cb94cdb11
proteore
parents:
12
diff
changeset
|
546 writer = csv.writer(output,delimiter="\t") |
|
12
f6afaa1f562c
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents:
11
diff
changeset
|
547 |
|
6
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
548 nextprot_file=[["NextprotID","MW","SeqLength","IsoPoint","Chr","SubcellLocations","Diseases","TMDomains","ProteinExistence"]] |
|
14
3f9ba522bfe8
planemo upload commit 272d182247f4fa55a928ec8cdcb83eab50c7f6ef
proteore
parents:
13
diff
changeset
|
549 writer.writerows(nextprot_file) |
|
3f9ba522bfe8
planemo upload commit 272d182247f4fa55a928ec8cdcb83eab50c7f6ef
proteore
parents:
13
diff
changeset
|
550 |
|
6
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
551 for id in ids : |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
552 #print (id) |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
553 query="https://api.nextprot.org/entry/"+id+".json" |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
554 resp = requests.get(url=query) |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
555 data = resp.json() |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
556 |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
557 #get info from json dictionary |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
558 mass_mol = data["entry"]["isoforms"][0]["massAsString"] |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
559 seq_length = data['entry']["isoforms"][0]["sequenceLength"] |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
560 iso_elec_point = data['entry']["isoforms"][0]["isoelectricPointAsString"] |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
561 chr_loc = data['entry']["chromosomalLocations"][0]["chromosome"] |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
562 protein_existence = "PE"+str(data['entry']["overview"]['proteinExistence']['level']) |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
563 |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
564 #put all subcell loc in a set |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
565 if "subcellular-location" in data['entry']["annotationsByCategory"].keys() : |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
566 subcell_locs = data['entry']["annotationsByCategory"]["subcellular-location"] |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
567 all_subcell_locs = set() |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
568 for loc in subcell_locs : |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
569 all_subcell_locs.add(loc['cvTermName']) |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
570 all_subcell_locs.discard("") |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
571 all_subcell_locs = ";".join(all_subcell_locs) |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
572 else : |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
573 all_subcell_locs = "NA" |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
574 |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
575 #put all subcell loc in a set |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
576 if ('disease') in data['entry']['annotationsByCategory'].keys() : |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
577 diseases = data['entry']['annotationsByCategory']['disease'] |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
578 all_diseases = set() |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
579 for disease in diseases : |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
580 if (disease['cvTermName'] is not None and disease['cvTermName'] != ""): |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
581 all_diseases.add(disease['cvTermName']) |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
582 if len(all_diseases) > 0 : all_diseases = ";".join(all_diseases) |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
583 else : all_diseases="NA" |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
584 else : |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
585 all_diseases="NA" |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
586 |
|
11
ac2cd728c40e
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents:
9
diff
changeset
|
587 #get all tm domain |
|
6
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
588 nb_domains = 0 |
|
11
ac2cd728c40e
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents:
9
diff
changeset
|
589 if "transmembrane-region" in data['entry']['annotationsByCategory'].keys(): |
|
ac2cd728c40e
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents:
9
diff
changeset
|
590 tm_domains = data['entry']['annotationsByCategory']["transmembrane-region"] |
|
ac2cd728c40e
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents:
9
diff
changeset
|
591 all_tm_domains = set() |
|
ac2cd728c40e
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents:
9
diff
changeset
|
592 for tm in tm_domains : |
|
ac2cd728c40e
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents:
9
diff
changeset
|
593 all_tm_domains.add(tm['cvTermName']) |
|
ac2cd728c40e
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents:
9
diff
changeset
|
594 nb_domains+=1 |
|
24
6cbb76823b7b
planemo upload commit 3e8aad84b7fe9218370364b1aef3ea5779475f9f-dirty
proteore
parents:
23
diff
changeset
|
595 #print "nb domains ++" |
|
6cbb76823b7b
planemo upload commit 3e8aad84b7fe9218370364b1aef3ea5779475f9f-dirty
proteore
parents:
23
diff
changeset
|
596 #print (nb_domains) |
|
15
5f39d3ef5fe3
planemo upload commit 45108c8cbeeccab12031bbf6e65f1177222d4e10
proteore
parents:
14
diff
changeset
|
597 nextprot_file[:] = [] |
|
12
f6afaa1f562c
planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c-dirty
proteore
parents:
11
diff
changeset
|
598 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
|
599 writer.writerows(nextprot_file) |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
600 |
|
24
6cbb76823b7b
planemo upload commit 3e8aad84b7fe9218370364b1aef3ea5779475f9f-dirty
proteore
parents:
23
diff
changeset
|
601 id = str(10000000000 - int(time.strftime("%Y%m%d"))) |
|
6cbb76823b7b
planemo upload commit 3e8aad84b7fe9218370364b1aef3ea5779475f9f-dirty
proteore
parents:
23
diff
changeset
|
602 |
|
6cbb76823b7b
planemo upload commit 3e8aad84b7fe9218370364b1aef3ea5779475f9f-dirty
proteore
parents:
23
diff
changeset
|
603 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
|
604 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_nextprot_ref") |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
605 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
606 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
607 # Main function |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
608 ####################################################################################################### |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
609 def main(): |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
610 parser = argparse.ArgumentParser() |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
611 parser.add_argument("--hpa", metavar = ("HPA_OPTION")) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
612 parser.add_argument("--peptideatlas", metavar=("SAMPLE_CATEGORY_ID")) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
613 parser.add_argument("--id_mapping", metavar = ("ID_MAPPING_SPECIES")) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
614 parser.add_argument("--interactome", metavar = ("PPI")) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
615 parser.add_argument("--species") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
616 parser.add_argument("--date") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
617 parser.add_argument("-o", "--output") |
|
6
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
618 parser.add_argument("--database") |
|
31
faeeabb11a4d
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
30
diff
changeset
|
619 parser.add_argument("--tool_data_path") |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
620 args = parser.parse_args() |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
621 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
622 data_manager_dict = {} |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
623 # Extract json file params |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
624 filename = args.output |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
625 params = from_json_string(open(filename).read()) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
626 target_directory = params[ 'output_data' ][0]['extra_files_path'] |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
627 os.mkdir(target_directory) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
628 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
629 ## Download source files from HPA |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
630 try: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
631 hpa = args.hpa |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
632 except NameError: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
633 hpa = None |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
634 if hpa is not None: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
635 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
636 hpa = hpa.split(",") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
637 for hpa_tissue in hpa: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
638 HPA_sources(data_manager_dict, hpa_tissue, target_directory) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
639 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
640 ## Download source file from Peptide Atlas query |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
641 try: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
642 peptide_atlas = args.peptideatlas |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
643 date = args.date |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
644 except NameError: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
645 peptide_atlas = None |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
646 if peptide_atlas is not None: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
647 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
648 peptide_atlas = peptide_atlas.split(",") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
649 for pa_tissue in peptide_atlas: |
|
32
ec1febc6672e
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
31
diff
changeset
|
650 peptide_atlas_sources(data_manager_dict, pa_tissue, date, target_directory) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
651 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
652 ## Download ID_mapping source file from Uniprot |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
653 try: |
|
40
a7bcc9a0a456
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
39
diff
changeset
|
654 id_mapping = args.id_mapping |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
655 except NameError: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
656 id_mapping = None |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
657 if id_mapping is not None: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
658 id_mapping = id_mapping .split(",") |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
659 for species in id_mapping : |
|
33
e681d7de4f5e
"planemo upload commit ba867b8fa3352695fbda1ae764407f363ee79a50-dirty"
proteore
parents:
32
diff
changeset
|
660 id_mapping_sources(data_manager_dict, species, target_directory, args.tool_data_path) |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
661 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
662 ## Download PPI ref files from biogrid/bioplex/humap |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
663 try: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
664 interactome=args.interactome |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
665 if interactome == "biogrid" : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
666 species=args.species |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
667 else : |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
668 species="Human" |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
669 except NameError: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
670 interactome=None |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
671 species=None |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
672 if interactome is not None and species is not None: |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
673 PPI_ref_files(data_manager_dict, species, interactome, target_directory) |
|
6
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
674 |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
675 ## Build nextprot ref file for add protein features |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
676 try: |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
677 database=args.database |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
678 except NameError: |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
679 database=None |
|
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
680 if database is not None : |
|
8
d5badf9de1b0
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
6
diff
changeset
|
681 Build_nextprot_ref_file(data_manager_dict,target_directory) |
|
6
f281a1eb83d6
planemo upload commit 5df487b88ce2146c4be8e1d9f419006583185f6a-dirty
proteore
parents:
5
diff
changeset
|
682 |
|
0
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
683 #save info to json file |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
684 filename = args.output |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
685 open(filename, 'wb').write(to_json_string(data_manager_dict)) |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
686 |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
687 if __name__ == "__main__": |
|
0a26460d7366
planemo upload commit dbc027f59706f5b7d3f9f9319f2652baa50e2df5-dirty
proteore
parents:
diff
changeset
|
688 main() |
