Mercurial > repos > dchristiany > data_manager_proteore
annotate data_manager/resource_building.py @ 44:f89f008bf80d draft
planemo upload commit 9b701c1faf8be4835b4e7236780ee9ee26f9a373-dirty
author | dchristiany |
---|---|
date | Thu, 31 Jan 2019 08:54:10 -0500 |
parents | 3febf3d1139a |
children | 80fc0b28e227 |
rev | line source |
---|---|
19
85532a48e4e4
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
17
diff
changeset
|
1 # -*- coding: utf-8 -*- |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
2 """ |
10
2f153b41b6fe
planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents:
7
diff
changeset
|
3 The purpose of this script is to create source files from different databases to be used in other proteore tools |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
4 """ |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
5 |
21
0a79066992fc
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
20
diff
changeset
|
6 import os, sys, argparse, requests, time, csv, re, json, shutil, zipfile |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
7 from io import BytesIO |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
8 from zipfile import ZipFile |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
9 from galaxy.util.json import from_json_string, to_json_string |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
10 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
11 ####################################################################################################### |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
12 # General functions |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
13 ####################################################################################################### |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
14 def unzip(url, output_file): |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
15 """ |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
16 Get a zip file content from a link and unzip |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
17 """ |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
18 content = requests.get(url) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
19 zipfile = ZipFile(BytesIO(content.content)) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
20 output_content = "" |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
21 output_content += zipfile.open(zipfile.namelist()[0]).read() |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
22 output = open(output_file, "w") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
23 output.write(output_content) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
24 output.close() |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
25 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
26 def _add_data_table_entry(data_manager_dict, data_table_entry,data_table): |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
27 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {}) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
28 data_manager_dict['data_tables'][data_table] = data_manager_dict['data_tables'].get(data_table, []) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
29 data_manager_dict['data_tables'][data_table].append(data_table_entry) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
30 return data_manager_dict |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
31 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
32 ####################################################################################################### |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
33 # 1. Human Protein Atlas |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
34 # - Normal tissue |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
35 # - Pathology |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
36 # - Full Atlas |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
37 ####################################################################################################### |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
38 def HPA_sources(data_manager_dict, tissue, target_directory): |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
39 if tissue == "HPA_normal_tissue": |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
40 tissue_name = "HPA normal tissue" |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
41 url = "https://www.proteinatlas.org/download/normal_tissue.tsv.zip" |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
42 elif tissue == "HPA_pathology": |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
43 tissue_name = "HPA pathology" |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
44 url = "https://www.proteinatlas.org/download/pathology.tsv.zip" |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
45 elif tissue == "HPA_full_atlas": |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
46 tissue_name = "HPA full atlas" |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
47 url = "https://www.proteinatlas.org/download/proteinatlas.tsv.zip" |
10
2f153b41b6fe
planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents:
7
diff
changeset
|
48 |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
49 output_file = tissue +"_"+ time.strftime("%d-%m-%Y") + ".tsv" |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
50 path = os.path.join(target_directory, output_file) |
10
2f153b41b6fe
planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents:
7
diff
changeset
|
51 unzip(url, path) #download and save file |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
52 tissue_name = tissue_name + " " + time.strftime("%d/%m/%Y") |
10
2f153b41b6fe
planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents:
7
diff
changeset
|
53 tissue_id = tissue_name.replace(" ","_").replace("/","-") |
2f153b41b6fe
planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents:
7
diff
changeset
|
54 |
2f153b41b6fe
planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents:
7
diff
changeset
|
55 data_table_entry = dict(id=tissue_id, name = tissue_name, value = tissue, path = path) |
12
60cb0a5ae661
planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents:
10
diff
changeset
|
56 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_protein_atlas") |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
57 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
58 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
59 ####################################################################################################### |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
60 # 2. Peptide Atlas |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
61 ####################################################################################################### |
43
3febf3d1139a
planemo upload commit 9b701c1faf8be4835b4e7236780ee9ee26f9a373-dirty
dchristiany
parents:
42
diff
changeset
|
62 def peptide_atlas_sources(data_manager_dict, tissue, date, target_directory): |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
63 # Define organism_id (here Human) - to be upraded when other organism added to the project |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
64 organism_id = "2" |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
65 # Extract sample_category_id and output filename |
34
0c0586ac3e29
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
33
diff
changeset
|
66 tissue=tissue.split(".") |
29
871a7347ca24
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
28
diff
changeset
|
67 sample_category_id = tissue[0] |
34
0c0586ac3e29
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
33
diff
changeset
|
68 tissue_name = tissue[1] |
43
3febf3d1139a
planemo upload commit 9b701c1faf8be4835b4e7236780ee9ee26f9a373-dirty
dchristiany
parents:
42
diff
changeset
|
69 output_file = tissue_name+"_"+date + ".tsv" |
30
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
70 |
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
71 query="https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id="+ \ |
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
72 sample_category_id+"&display_options=ShowAbundances&organism_id="+organism_id+ \ |
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
73 "&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf\ |
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
74 &QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY" |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
75 |
30
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
76 with requests.Session() as s: |
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
77 download = s.get(query) |
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
78 decoded_content = download.content.decode('utf-8') |
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
79 cr = csv.reader(decoded_content.splitlines(), delimiter='\t') |
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
80 |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
81 uni_dict = build_dictionary(cr) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
82 |
10
2f153b41b6fe
planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents:
7
diff
changeset
|
83 #columns of data table peptide_atlas |
34
0c0586ac3e29
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
33
diff
changeset
|
84 tissue_id = tissue_name+"_"+date |
0c0586ac3e29
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
33
diff
changeset
|
85 name = tissue_id.replace("-","/").replace("_"," ") |
36
a4811c440b45
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
35
diff
changeset
|
86 path = os.path.join(target_directory,output_file) |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
87 |
30
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
88 with open(path,"w") as out : |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
89 w = csv.writer(out,delimiter='\t') |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
90 w.writerow(["Uniprot_AC","nb_obs"]) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
91 w.writerows(uni_dict.items()) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
92 |
34
0c0586ac3e29
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
33
diff
changeset
|
93 data_table_entry = dict(id=tissue_id, name=name, value = path, tissue = tissue_name) |
12
60cb0a5ae661
planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents:
10
diff
changeset
|
94 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_peptide_atlas") |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
95 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
96 #function to count the number of observations by uniprot id |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
97 def build_dictionary (csv) : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
98 uni_dict = {} |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
99 for line in csv : |
30
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
100 if "-" not in line[0] and check_uniprot_access(line[0]) : |
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
101 if line[0] in uni_dict : |
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
102 uni_dict[line[0]] += int(line[5]) |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
103 else : |
30
b8271b9a1049
planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty
dchristiany
parents:
29
diff
changeset
|
104 uni_dict[line[0]] = int(line[5]) |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
105 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
106 return uni_dict |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
107 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
108 #function to check if an id is an uniprot accession number : return True or False- |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
109 def check_uniprot_access (id) : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
110 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}") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
111 if uniprot_pattern.match(id) : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
112 return True |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
113 else : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
114 return False |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
115 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
116 ####################################################################################################### |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
117 # 3. ID mapping file |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
118 ####################################################################################################### |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
119 import ftplib, gzip |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
120 csv.field_size_limit(sys.maxsize) # to handle big files |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
121 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
122 def id_mapping_sources (data_manager_dict, species, target_directory) : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
123 |
28
d235909789ca
planemo upload commit e694c4b0df30a4286ba09721696e8ec3af25fd97-dirty
dchristiany
parents:
27
diff
changeset
|
124 human = species == "Human" |
d235909789ca
planemo upload commit e694c4b0df30a4286ba09721696e8ec3af25fd97-dirty
dchristiany
parents:
27
diff
changeset
|
125 species_dict = { "Human" : "HUMAN_9606", "Mouse" : "MOUSE_10090", "Rat" : "RAT_10116" } |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
126 files=["idmapping_selected.tab.gz","idmapping.dat.gz"] |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
127 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
128 #header |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
129 if human : tab = [["UniProt-AC","UniProt-ID","GeneID","RefSeq","GI","PDB","GO","PIR","MIM","UniGene","Ensembl_Gene","Ensembl_Transcript","Ensembl_Protein","neXtProt","BioGrid","STRING","KEGG"]] |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
130 else : tab = [["UniProt-AC","UniProt-ID","GeneID","RefSeq","GI","PDB","GO","PIR","MIM","UniGene","Ensembl_Gene","Ensembl_Transcript","Ensembl_Protein","BioGrid","STRING","KEGG"]] |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
131 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
132 #print("header ok") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
133 |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
134 #get selected.tab and keep only ids of interest |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
135 selected_tab_file=species_dict[species]+"_"+files[0] |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
136 tab_path = download_from_uniprot_ftp(selected_tab_file,target_directory) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
137 with gzip.open(tab_path,"rt") as select : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
138 tab_reader = csv.reader(select,delimiter="\t") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
139 for line in tab_reader : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
140 tab.append([line[i] for i in [0,1,2,3,4,5,6,11,13,14,18,19,20]]) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
141 os.remove(tab_path) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
142 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
143 #print("selected_tab ok") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
144 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
145 """ |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
146 Supplementary ID to get from HUMAN_9606_idmapping.dat : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
147 -NextProt,BioGrid,STRING,KEGG |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
148 """ |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
149 |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
150 #there's more id type for human |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
151 if human : ids = ['neXtProt','BioGrid','STRING','KEGG' ] #ids to get from dat_file |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
152 else : ids = ['BioGrid','STRING','KEGG' ] |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
153 unidict = {} |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
154 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
155 #keep only ids of interest in dictionaries |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
156 dat_file=species_dict[species]+"_"+files[1] |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
157 dat_path = download_from_uniprot_ftp(dat_file,target_directory) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
158 with gzip.open(dat_path,"rt") as dat : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
159 dat_reader = csv.reader(dat,delimiter="\t") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
160 for line in dat_reader : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
161 uniprotID=line[0] #UniProtID as key |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
162 id_type=line[1] #ID type of corresponding id, key of sub-dictionnary |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
163 cor_id=line[2] #corresponding id |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
164 if "-" not in id_type : #we don't keep isoform |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
165 if id_type in ids and uniprotID in unidict : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
166 if id_type in unidict[uniprotID] : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
167 unidict[uniprotID][id_type]= ";".join([unidict[uniprotID][id_type],cor_id]) #if there is already a value in the dictionnary |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
168 else : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
169 unidict[uniprotID].update({ id_type : cor_id }) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
170 elif id_type in ids : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
171 unidict[uniprotID]={id_type : cor_id} |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
172 os.remove(dat_path) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
173 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
174 #print("dat_file ok") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
175 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
176 #add ids from idmapping.dat to the final tab |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
177 for line in tab[1:] : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
178 uniprotID=line[0] |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
179 if human : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
180 if uniprotID in unidict : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
181 nextprot = access_dictionary(unidict,uniprotID,'neXtProt') |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
182 if nextprot != '' : nextprot = clean_nextprot_id(nextprot,line[0]) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
183 line.extend([nextprot,access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'), |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
184 access_dictionary(unidict,uniprotID,'KEGG')]) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
185 else : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
186 line.extend(["","","",""]) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
187 else : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
188 if uniprotID in unidict : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
189 line.extend([access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'), |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
190 access_dictionary(unidict,uniprotID,'KEGG')]) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
191 else : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
192 line.extend(["","",""]) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
193 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
194 #print ("tab ok") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
195 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
196 #add missing nextprot ID for human |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
197 if human : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
198 #build next_dict |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
199 nextprot_ids = id_list_from_nextprot_ftp("nextprot_ac_list_all.txt",target_directory) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
200 next_dict = {} |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
201 for nextid in nextprot_ids : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
202 next_dict[nextid.replace("NX_","")] = nextid |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
203 os.remove(os.path.join(target_directory,"nextprot_ac_list_all.txt")) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
204 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
205 #add missing nextprot ID |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
206 for line in tab[1:] : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
207 uniprotID=line[0] |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
208 nextprotID=line[13] |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
209 if nextprotID == '' and uniprotID in next_dict : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
210 line[13]=next_dict[uniprotID] |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
211 |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
212 output_file = species+"_id_mapping_"+ time.strftime("%d-%m-%Y") + ".tsv" |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
213 path = os.path.join(target_directory,output_file) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
214 |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
215 with open(path,"w") as out : |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
216 w = csv.writer(out,delimiter='\t') |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
217 w.writerows(tab) |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
218 |
28
d235909789ca
planemo upload commit e694c4b0df30a4286ba09721696e8ec3af25fd97-dirty
dchristiany
parents:
27
diff
changeset
|
219 name_dict={"Human" : "Homo sapiens", "Mouse" : "Mus musculus", "Rat" : "Rattus norvegicus"} |
d235909789ca
planemo upload commit e694c4b0df30a4286ba09721696e8ec3af25fd97-dirty
dchristiany
parents:
27
diff
changeset
|
220 name = species +" (" + name_dict[species]+" "+time.strftime("%d/%m/%Y")+")" |
10
2f153b41b6fe
planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents:
7
diff
changeset
|
221 id = species+"_id_mapping_"+ time.strftime("%d-%m-%Y") |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
222 |
10
2f153b41b6fe
planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents:
7
diff
changeset
|
223 data_table_entry = dict(id=id, name = name, value = species, path = path) |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
224 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_id_mapping") |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
225 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
226 def download_from_uniprot_ftp(file,target_directory) : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
227 ftp_dir = "pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/" |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
228 path = os.path.join(target_directory, file) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
229 ftp = ftplib.FTP("ftp.uniprot.org") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
230 ftp.login("anonymous", "anonymous") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
231 ftp.cwd(ftp_dir) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
232 ftp.retrbinary("RETR " + file, open(path, 'wb').write) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
233 ftp.quit() |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
234 return (path) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
235 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
236 def id_list_from_nextprot_ftp(file,target_directory) : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
237 ftp_dir = "pub/current_release/ac_lists/" |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
238 path = os.path.join(target_directory, file) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
239 ftp = ftplib.FTP("ftp.nextprot.org") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
240 ftp.login("anonymous", "anonymous") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
241 ftp.cwd(ftp_dir) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
242 ftp.retrbinary("RETR " + file, open(path, 'wb').write) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
243 ftp.quit() |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
244 with open(path,'r') as nextprot_ids : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
245 nextprot_ids = nextprot_ids.read().splitlines() |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
246 return (nextprot_ids) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
247 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
248 #return '' if there's no value in a dictionary, avoid error |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
249 def access_dictionary (dico,key1,key2) : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
250 if key1 in dico : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
251 if key2 in dico[key1] : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
252 return (dico[key1][key2]) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
253 else : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
254 return ("") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
255 #print (key2,"not in ",dico,"[",key1,"]") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
256 else : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
257 return ('') |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
258 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
259 #if there are several nextprot ID for one uniprotID, return the uniprot like ID |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
260 def clean_nextprot_id (next_id,uniprotAc) : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
261 if len(next_id.split(";")) > 1 : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
262 tmp = next_id.split(";") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
263 if "NX_"+uniprotAc in tmp : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
264 return ("NX_"+uniprotAc) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
265 else : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
266 return (tmp[1]) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
267 else : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
268 return (next_id) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
269 |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
270 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
271 ####################################################################################################### |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
272 # 4. Build protein interaction maps files |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
273 ####################################################################################################### |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
274 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
275 def PPI_ref_files(data_manager_dict, species, interactome, target_directory): |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
276 |
28
d235909789ca
planemo upload commit e694c4b0df30a4286ba09721696e8ec3af25fd97-dirty
dchristiany
parents:
27
diff
changeset
|
277 species_dict={'Human':'Homo sapiens',"Mouse":"Mus musculus","Rat":"Rattus norvegicus"} |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
278 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
279 ##BioGRID |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
280 if interactome=="biogrid": |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
281 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
282 tab2_link="https://downloads.thebiogrid.org/Download/BioGRID/Release-Archive/BIOGRID-3.5.167/BIOGRID-ORGANISM-3.5.167.tab2.zip" |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
283 |
39
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
284 #download zip file |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
285 r = requests.get(tab2_link) |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
286 with open("BioGRID.zip", "wb") as code: |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
287 code.write(r.content) |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
288 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
289 #unzip files |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
290 with zipfile.ZipFile("BioGRID.zip", 'r') as zip_ref: |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
291 if not os.path.exists("tmp_BioGRID"): os.makedirs("tmp_BioGRID") |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
292 zip_ref.extractall("tmp_BioGRID") |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
293 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
294 #import file of interest and build dictionary |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
295 file_path="tmp_BioGRID/BIOGRID-ORGANISM-"+species_dict[species].replace(" ","_")+"-3.5.167.tab2.txt" |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
296 with open(file_path,"r") as handle : |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
297 tab_file = csv.reader(handle,delimiter="\t") |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
298 dico_network = {} |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
299 GeneID_index=1 |
39
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
300 network_cols=[1,2,7,8,11,12,14,18,20] |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
301 for line in tab_file : |
22
778cc97cb115
planemo upload commit e2e4ac8c2c4dc5675d4c415ff192a925ca5e6b98-dirty
dchristiany
parents:
21
diff
changeset
|
302 if line[GeneID_index] not in dico_network: |
778cc97cb115
planemo upload commit e2e4ac8c2c4dc5675d4c415ff192a925ca5e6b98-dirty
dchristiany
parents:
21
diff
changeset
|
303 dico_network[line[GeneID_index]]=[[line[i] for i in network_cols]] |
778cc97cb115
planemo upload commit e2e4ac8c2c4dc5675d4c415ff192a925ca5e6b98-dirty
dchristiany
parents:
21
diff
changeset
|
304 else: |
778cc97cb115
planemo upload commit e2e4ac8c2c4dc5675d4c415ff192a925ca5e6b98-dirty
dchristiany
parents:
21
diff
changeset
|
305 dico_network[line[GeneID_index]].append([line[i] for i in network_cols]) |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
306 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
307 #delete tmp_BioGRID directory |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
308 os.remove("BioGRID.zip") |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
309 shutil.rmtree("tmp_BioGRID", ignore_errors=True) |
39
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
310 |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
311 #download NCBI2Reactome.txt file and build dictionary |
39
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
312 with requests.Session() as s: |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
313 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt') |
40
fddf4a3847f4
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
39
diff
changeset
|
314 r.encoding ="utf-8" |
fddf4a3847f4
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
39
diff
changeset
|
315 tab_file = csv.reader(r.content.splitlines(), delimiter='\t') |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
316 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
317 dico_nodes = {} |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
318 uniProt_index=0 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
319 pathway_description_index=3 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
320 species_index=5 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
321 for line in tab_file : |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
322 if line[species_index]==species_dict[species]: |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
323 if line[uniProt_index] in dico_nodes : |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
324 dico_nodes[line[uniProt_index]].append(line[pathway_description_index]) |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
325 else : |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
326 dico_nodes[line[uniProt_index]] = [line[pathway_description_index]] |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
327 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
328 dico={} |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
329 dico['network']=dico_network |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
330 dico['nodes']=dico_nodes |
39
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
331 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
332 ##Bioplex |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
333 elif interactome=="bioplex": |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
334 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
335 with requests.Session() as s: |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
336 r = s.get('http://bioplex.hms.harvard.edu/data/BioPlex_interactionList_v4a.tsv') |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
337 r = r.content.decode('utf-8') |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
338 bioplex = csv.reader(r.splitlines(), delimiter='\t') |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
339 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
340 dico_network = {} |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
341 dico_network["GeneID"]={} |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
342 network_geneid_cols=[0,1,4,5,8] |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
343 dico_network["UniProt-AC"]={} |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
344 network_uniprot_cols=[2,3,4,5,8] |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
345 dico_GeneID_to_UniProt = {} |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
346 for line in bioplex : |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
347 if line[0] not in dico_network["GeneID"]: |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
348 dico_network["GeneID"][line[0]]=[[line[i] for i in network_geneid_cols]] |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
349 else : |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
350 dico_network["GeneID"][line[0]].append([line[i] for i in network_geneid_cols]) |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
351 if line[1] not in dico_network["UniProt-AC"]: |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
352 dico_network["UniProt-AC"][line[2]]=[[line[i] for i in network_uniprot_cols]] |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
353 else: |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
354 dico_network["UniProt-AC"][line[2]].append([line[i] for i in network_uniprot_cols]) |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
355 dico_GeneID_to_UniProt[line[0]]=line[2] |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
356 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
357 with requests.Session() as s: |
40
fddf4a3847f4
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
39
diff
changeset
|
358 r = s.get('https://reactome.org/download/current/UniProt2Reactome.txt') |
fddf4a3847f4
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
39
diff
changeset
|
359 r.encoding ="utf-8" |
fddf4a3847f4
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
39
diff
changeset
|
360 tab_file = csv.reader(r.content.splitlines(), delimiter='\t') |
39
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
361 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
362 dico_nodes_uniprot = {} |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
363 uniProt_index=0 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
364 pathway_description_index=3 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
365 species_index=5 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
366 for line in tab_file : |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
367 if line[species_index]==species_dict[species]: |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
368 if line[uniProt_index] in dico_nodes_uniprot : |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
369 dico_nodes_uniprot[line[uniProt_index]].append(line[pathway_description_index]) |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
370 else : |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
371 dico_nodes_uniprot[line[uniProt_index]] = [line[pathway_description_index]] |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
372 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
373 with requests.Session() as s: |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
374 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt') |
40
fddf4a3847f4
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
39
diff
changeset
|
375 r.encoding ="utf-8" |
fddf4a3847f4
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
39
diff
changeset
|
376 tab_file = csv.reader(r.content.splitlines(), delimiter='\t') |
39
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
377 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
378 dico_nodes_geneid = {} |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
379 uniProt_index=0 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
380 pathway_description_index=3 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
381 species_index=5 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
382 for line in tab_file : |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
383 if line[species_index]==species_dict[species]: |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
384 if line[uniProt_index] in dico_nodes_geneid : |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
385 dico_nodes_geneid[line[uniProt_index]].append(line[pathway_description_index]) |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
386 else : |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
387 dico_nodes_geneid[line[uniProt_index]] = [line[pathway_description_index]] |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
388 |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
389 dico={} |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
390 dico_nodes={} |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
391 dico_nodes['GeneID']=dico_nodes_geneid |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
392 dico_nodes['UniProt-AC']=dico_nodes_uniprot |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
393 dico['network']=dico_network |
ec6252ad1a8e
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
36
diff
changeset
|
394 dico['nodes']=dico_nodes |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
395 dico['convert']=dico_GeneID_to_UniProt |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
396 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
397 #writing output |
42
5a37a086c9a8
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
41
diff
changeset
|
398 output_file = species+'_'+interactome+'_'+ time.strftime("%d-%m-%Y") + ".json" |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
399 path = os.path.join(target_directory,output_file) |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
400 name = species+" ("+species_dict[species]+") "+time.strftime("%d/%m/%Y") |
27
592c59530c32
planemo upload commit e694c4b0df30a4286ba09721696e8ec3af25fd97-dirty
dchristiany
parents:
26
diff
changeset
|
401 id = species+"_"+interactome+"_"+ time.strftime("%d-%m-%Y") |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
402 |
41
4062484f2cee
planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty
dchristiany
parents:
40
diff
changeset
|
403 with open(path, 'w') as handle: |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
404 json.dump(dico, handle, sort_keys=True) |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
405 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
406 data_table_entry = dict(id=id, name = name, value = species, path = path) |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
407 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_"+interactome+"_dictionaries") |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
408 |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
409 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
410 ####################################################################################################### |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
411 # Main function |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
412 ####################################################################################################### |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
413 def main(): |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
414 parser = argparse.ArgumentParser() |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
415 parser.add_argument("--hpa", metavar = ("HPA_OPTION")) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
416 parser.add_argument("--peptideatlas", metavar=("SAMPLE_CATEGORY_ID")) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
417 parser.add_argument("--id_mapping", metavar = ("ID_MAPPING_SPECIES")) |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
418 parser.add_argument("--interactome", metavar = ("PPI")) |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
419 parser.add_argument("--species") |
43
3febf3d1139a
planemo upload commit 9b701c1faf8be4835b4e7236780ee9ee26f9a373-dirty
dchristiany
parents:
42
diff
changeset
|
420 parser.add_argument("--date") |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
421 parser.add_argument("-o", "--output") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
422 args = parser.parse_args() |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
423 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
424 data_manager_dict = {} |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
425 # Extract json file params |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
426 filename = args.output |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
427 params = from_json_string(open(filename).read()) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
428 target_directory = params[ 'output_data' ][0]['extra_files_path'] |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
429 os.mkdir(target_directory) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
430 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
431 ## Download source files from HPA |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
432 try: |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
433 hpa = args.hpa |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
434 except NameError: |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
435 hpa = None |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
436 if hpa is not None: |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
437 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/" |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
438 hpa = hpa.split(",") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
439 for hpa_tissue in hpa: |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
440 HPA_sources(data_manager_dict, hpa_tissue, target_directory) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
441 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
442 ## Download source file from Peptide Atlas query |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
443 try: |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
444 peptide_atlas = args.peptideatlas |
43
3febf3d1139a
planemo upload commit 9b701c1faf8be4835b4e7236780ee9ee26f9a373-dirty
dchristiany
parents:
42
diff
changeset
|
445 date = args.date |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
446 except NameError: |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
447 peptide_atlas = None |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
448 if peptide_atlas is not None: |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
449 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/" |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
450 peptide_atlas = peptide_atlas.split(",") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
451 for pa_tissue in peptide_atlas: |
43
3febf3d1139a
planemo upload commit 9b701c1faf8be4835b4e7236780ee9ee26f9a373-dirty
dchristiany
parents:
42
diff
changeset
|
452 peptide_atlas_sources(data_manager_dict, pa_tissue, date, target_directory) |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
453 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
454 ## Download ID_mapping source file from Uniprot |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
455 try: |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
456 id_mapping=args.id_mapping |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
457 except NameError: |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
458 id_mapping = None |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
459 if id_mapping is not None: |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
460 id_mapping = id_mapping .split(",") |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
461 for species in id_mapping : |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
462 id_mapping_sources(data_manager_dict, species, target_directory) |
15
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
463 |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
464 ## Download PPI ref files from biogrid/bioplex/humap |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
465 try: |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
466 interactome=args.interactome |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
467 species=args.species |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
468 except NameError: |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
469 interactome=None |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
470 species=None |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
471 if interactome is not None and species is not None: |
83f57ba70416
planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty
dchristiany
parents:
13
diff
changeset
|
472 PPI_ref_files(data_manager_dict, species, interactome, target_directory) |
0
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
473 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
474 #save info to json file |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
475 filename = args.output |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
476 open(filename, 'wb').write(to_json_string(data_manager_dict)) |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
477 |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
478 if __name__ == "__main__": |
2de84fea8367
planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff
changeset
|
479 main() |