Mercurial > repos > dchristiany > frogs_data_manager
annotate data_manager/FROGS_data_manager.py @ 9:0d9cb5c5aa35 draft
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
author | dchristiany |
---|---|
date | Mon, 07 Oct 2019 10:56:23 -0400 |
parents | 99b2dfb1fa1c |
children | 238a5328279d |
rev | line source |
---|---|
0
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
1 # -*- coding: utf-8 -*- |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
2 from galaxy.util.json import from_json_string, to_json_string |
5
ac4fb9d97a51
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
4
diff
changeset
|
3 import os, sys, argparse, time, json, requests, urllib, tarfile |
0
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
4 |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
5 def get_args(): |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
6 parser = argparse.ArgumentParser() |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
7 parser.add_argument("-d","--database") |
9
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
8 parser.add_argument("--custom_db") |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
9 parser.add_argument("--amplicons") |
0
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
10 parser.add_argument("-o","--output") |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
11 args = parser.parse_args() |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
12 return args |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
13 |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
14 def _add_data_table_entry(data_manager_dict, data_table_entry,data_table): |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
15 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {}) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
16 data_manager_dict['data_tables'][data_table] = data_manager_dict['data_tables'].get(data_table, []) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
17 data_manager_dict['data_tables'][data_table].append(data_table_entry) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
18 return data_manager_dict |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
19 |
9
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
20 def frogs_sources(data_manager_dict,target_directory,amplicons_list): |
0
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
21 |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
22 #get frogs database index |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
23 frogs_db_index_link="http://genoweb.toulouse.inra.fr/frogs_databanks/assignation/FROGS_databases.tsv" |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
24 with requests.Session() as s: |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
25 download = s.get(frogs_db_index_link) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
26 decoded_content = download.content.decode('utf-8') |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
27 db_index = download.content.splitlines() |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
28 db_index = [line.split("\t") for line in db_index[1:]] |
7
99b2dfb1fa1c
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
6
diff
changeset
|
29 db_index = [line[:4]+[line[1]+"_"+line[2]+"_"+line[3]]+[line[4]] for line in db_index] #add column name |
4
95d5191f1dc4
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
0
diff
changeset
|
30 |
9
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
31 #filter amplicons |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
32 if len(amplicons_list)!=0: |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
33 db_index = [line for line in db_index if line[4] in amplicons_list] |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
34 |
0
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
35 #get frogs dbs |
7
99b2dfb1fa1c
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
6
diff
changeset
|
36 os.chdir(target_directory) |
0
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
37 dir_name="frogs_db_"+time.strftime("%Y%m%d") |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
38 os.mkdir(dir_name) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
39 dbs=set([]) |
7
99b2dfb1fa1c
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
6
diff
changeset
|
40 for line in db_index: |
0
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
41 value=line[4] |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
42 name=value.replace("_"," ") |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
43 link=line[5] |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
44 |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
45 #download frogs db |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
46 dl_file = urllib.URLopener() |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
47 dl_file.retrieve(link, "tmp.tar.gz") |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
48 |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
49 #unzip frogs db |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
50 with tarfile.open("tmp.tar.gz") as tar: |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
51 tar.extractall(dir_name) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
52 tar.close() |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
53 os.remove('tmp.tar.gz') |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
54 |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
55 #get fasta file path |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
56 tmp = set(os.listdir(dir_name)) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
57 new_db = dir_name+"/"+"".join(tmp.difference(dbs)) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
58 files = os.listdir(new_db) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
59 fasta = "".join([file for file in files if file.endswith('.fasta')]) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
60 path = new_db+'/'+fasta |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
61 dbs = os.listdir(dir_name) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
62 release = value+"_"+time.strftime("%Y-%m-%d") |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
63 date=time.strftime("%Y%m%d") |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
64 path = os.path.join(target_directory,path) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
65 |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
66 data_table_entry = dict(name = name, value = value, path=path) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
67 _add_data_table_entry(data_manager_dict, data_table_entry, "frogs_db") |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
68 |
9
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
69 def HVL_sources(data_manager_dict,target_directory): |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
70 |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
71 #get phiX files |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
72 os.chdir(target_directory) |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
73 for link in ["http://genoweb.toulouse.inra.fr/frogs_databanks/HVL/ITS/UNITE_s_7.1_20112016/Unite_s_7.1_20112016_ITS1.fasta","http://genoweb.toulouse.inra.fr/frogs_databanks/HVL/ITS/UNITE_s_7.1_20112016/Unite_s_7.1_20112016_ITS2.fasta"]: |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
74 file_name=link.split("/")[-1].replace('.fasta',"_"+time.strftime("%Y-%m-%d")+".fasta") |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
75 dl_file = urllib.URLopener() |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
76 dl_file.retrieve(link,file_name) |
0
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
77 |
9
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
78 #get fasta file path |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
79 path = os.path.join(target_directory,file_name) |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
80 if link.endswith('ITS1.fasta'): |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
81 name = "UNITE 7.1 ITS1 " + time.strftime("%Y-%m-%d") |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
82 elif link.endswith('ITS2.fasta'): |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
83 name = "UNITE 7.1 ITS2 " + time.strftime("%Y-%m-%d") |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
84 value=file_name.replace('.fasta','') |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
85 |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
86 data_table_entry = dict(name = name, value = value, path=path) |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
87 _add_data_table_entry(data_manager_dict, data_table_entry, "HVL_db") |
0
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
88 |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
89 def main(): |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
90 |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
91 #get args from command line |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
92 args = get_args() |
9
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
93 if args.database=="frogs_db_data" and args.custom_db=="true": |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
94 amplicons_list = args.amplicons.split(",") |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
95 else : |
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
96 amplicons_list = [] |
0
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
97 |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
98 # Extract json file params |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
99 data_manager_dict = {} |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
100 filename = args.output |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
101 params = from_json_string(open(filename).read()) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
102 target_directory = params[ 'output_data' ][0]['extra_files_path'] |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
103 os.mkdir(target_directory) |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
104 |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
105 if args.database=="frogs_db_data": |
9
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
106 frogs_sources(data_manager_dict,target_directory,amplicons_list) |
0
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
107 elif args.database=="HVL_db_data": |
9
0d9cb5c5aa35
planemo upload commit af25bf8e6c89c9de2e25dc0bc0ead06d8c05cf17
dchristiany
parents:
7
diff
changeset
|
108 HVL_sources(data_manager_dict,target_directory) |
0
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
109 |
6
d11bc4a8f596
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
5
diff
changeset
|
110 #save info to json file |
d11bc4a8f596
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
5
diff
changeset
|
111 open(filename, 'wb').write(to_json_string(data_manager_dict)) |
d11bc4a8f596
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
5
diff
changeset
|
112 |
0
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
113 if __name__ == "__main__": |
7caea40b2a30
planemo upload commit c3ff1475af0e964a0c61458b66e2744c903d8d3d-dirty
dchristiany
parents:
diff
changeset
|
114 main() |