annotate data_manager/data_manager.py @ 3:c29275496061 draft

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit 8533fe71d1d50f09348da2dc34941724407a1ffe"
author iuc
date Tue, 14 Jul 2020 07:37:09 -0400
parents a6a81f921701
children 38ccd4589990
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
1 import argparse
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
2 import json
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
3 import os
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
4 try:
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
5 # For Python 3.0 and later
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
6 from urllib.request import Request, urlopen
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
7 except ImportError:
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
8 # Fall back to Python 2 imports
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
9 from urllib2 import Request, urlopen
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
10
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
11 DEFAULT_TAXLEVELS = "Kingdom,Phylum,Class,Order,Family,Genus,Species"
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
12
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
13 FILE2NAME = {
2
a6a81f921701 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit a096d3eb593edee6cb08247be3a09138d088a6ce"
iuc
parents: 1
diff changeset
14 "silva_138": "Silva version 138",
0
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
15 "silva_132": "Silva version 132",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
16 "silva_128": "Silva version 128",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
17 "rdp_16": "RDP trainset 16",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
18 "rdp_14": "RDP trainset 14",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
19 "greengenes_13.84": "GreenGenes version 13.84",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
20 "unite_8.0_fungi": "UNITE: General Fasta release 8.0 for Fungi",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
21 "unite_8.0_fungi_singletons": "UNITE: General Fasta release 8.0 for Fungi including global and 97% singletons",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
22 "RefSeq_RDP_2018_05": "NCBI RefSeq 16S rRNA database supplemented by RDP (05/2018)",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
23 "gtdb_2018_11": "GTDB: Genome Taxonomy Database (Bacteria & Archaea) (11/2018)",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
24 "hitdb_1": "HitDB version 1 (Human InTestinal 16S rRNA)",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
25 "silva_euk_18S_132": "Silva version 132 Eukaryotic 18S",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
26 "PR2_4.11.1": "Protist Ribosomal Reference database (PR2) 4.11.1"
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
27 }
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
28
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
29 FILE2TAXURL = {
2
a6a81f921701 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit a096d3eb593edee6cb08247be3a09138d088a6ce"
iuc
parents: 1
diff changeset
30 "silva_138": "https://zenodo.org/record/3731176/files/silva_nr_v138_train_set.fa.gz?download=1",
0
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
31 "silva_132": "https://zenodo.org/record/1172783/files/silva_nr_v132_train_set.fa.gz?download=1",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
32 "silva_128": "https://zenodo.org/record/824551/files/silva_nr_v128_train_set.fa.gz?download=1",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
33 "rdp_16": "https://zenodo.org/record/801828/files/rdp_train_set_16.fa.gz?download=1",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
34 "rdp_14": "https://zenodo.org/record/158955/files/rdp_train_set_14.fa.gz?download=1",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
35 "unite_8.0_fungi": "https://files.plutof.ut.ee/public/orig/EB/0C/EB0CCB3A871B77EA75E472D13926271076904A588D2E1C1EA5AFCF7397D48378.zip",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
36 "unite_8.0_fungi_singletons": "https://files.plutof.ut.ee/doi/06/A2/06A2C86256EED64085670EB0C54B7115F6DAC8F311C656A9CB33E386CFABA0D0.zip",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
37 "greengenes_13.84": "https://zenodo.org/record/158955/files/gg_13_8_train_set_97.fa.gz?download=1",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
38 "RefSeq_RDP_2018_05": "https://zenodo.org/record/2541239/files/RefSeq-RDP16S_v2_May2018.fa.gz?download=1",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
39 "gtdb_2018_11": "https://zenodo.org/record/2541239/files/GTDB_bac-arc_ssu_r86.fa.gz?download=1",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
40 "hitdb_1": "https://zenodo.org/record/159205/files/hitdb_v1.00.fa.gz?download=1",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
41 "silva_euk_18S_132": "https://zenodo.org/record/1447330/files/silva_132.18s.99_rep_set.dada2.fa.gz?download=1",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
42 "PR2_4.11.1": "https://github.com/pr2database/pr2database/releases/download/4.11.1/pr2_version_4.11.1_dada2.fasta.gz"
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
43 }
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
44
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
45 FILE2SPECIESURL = {
2
a6a81f921701 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit a096d3eb593edee6cb08247be3a09138d088a6ce"
iuc
parents: 1
diff changeset
46 "silva_138": "https://zenodo.org/record/3731176/files/silva_species_assignment_v138.fa.gz?download=1",
0
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
47 "silva_132": "https://zenodo.org/record/1172783/files/silva_species_assignment_v132.fa.gz?download=1",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
48 "silva_128": "https://zenodo.org/record/824551/files/silva_species_assignment_v128.fa.gz?download=1",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
49 "rdp_16": "https://zenodo.org/record/801828/files/rdp_species_assignment_16.fa.gz?download=1",
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
50 "rdp_14": "https://zenodo.org/record/158955/files/rdp_species_assignment_14.fa.gz?download=1"
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
51 }
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
52
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
53 FILE2TAXLEVELS = {
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
54 "PR2_4.11.1": "Kingdom,Supergroup,Division,Class,Order,Family,Genus,Species"
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
55 }
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
56
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
57
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
58 def url_download(url, fname, workdir):
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
59 """
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
60 download url to workdir/fname
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
61 """
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
62 file_path = os.path.join(workdir, fname)
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
63 if not os.path.exists(workdir):
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
64 os.makedirs(workdir)
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
65 src = None
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
66 dst = None
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
67 try:
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
68 req = Request(url)
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
69 src = urlopen(req)
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
70 with open(file_path, 'wb') as dst:
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
71 while True:
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
72 chunk = src.read(2**10)
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
73 if chunk:
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
74 dst.write(chunk)
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
75 else:
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
76 break
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
77 finally:
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
78 if src:
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
79 src.close()
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
80
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
81 # special treatment of UNITE DBs: they are zip files containing two fasta (xyz.fasta and developer/xyz.fasta)
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
82 if fname.startswith("unite"):
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
83 import glob
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
84 import gzip
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
85 import shutil
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
86 import zipfile
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
87 # unzip download
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
88 zip_ref = zipfile.ZipFile(file_path, 'r')
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
89 zip_ref.extractall(workdir)
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
90 zip_ref.close()
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
91 # gzip top level fasta file
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
92 fastas = glob.glob("%s/*fasta" % workdir)
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
93 if len(fastas) != 1:
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
94 msg = "UNITE download %s contained %d fasta file(s): %s" % (url, len(fastas), " ".join(fastas))
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
95 raise Exception(msg)
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
96 with open(fastas[0], 'rb') as f_in:
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
97 with gzip.open(file_path, 'wb') as f_out:
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
98 shutil.copyfileobj(f_in, f_out)
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
99
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
100
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
101 def remote_dataset(dataset, outjson):
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
102
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
103 with open(outjson) as jf:
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
104 params = json.loads(jf.read())
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
105
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
106 workdir = params['output_data'][0]['extra_files_path']
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
107 os.mkdir(workdir)
3
c29275496061 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit 8533fe71d1d50f09348da2dc34941724407a1ffe"
iuc
parents: 2
diff changeset
108 url_download(FILE2TAXURL[dataset], dataset + ".taxonomy", workdir)
0
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
109
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
110 data_manager_json = {"data_tables": {}}
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
111 data_manager_entry = {}
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
112 data_manager_entry['value'] = dataset
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
113 data_manager_entry['name'] = FILE2NAME[dataset]
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
114 data_manager_entry['path'] = dataset + ".taxonomy"
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
115 data_manager_entry['taxlevels'] = FILE2TAXLEVELS.get(dataset, DEFAULT_TAXLEVELS)
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
116 data_manager_json["data_tables"]["dada2_taxonomy"] = data_manager_entry
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
117
3
c29275496061 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit 8533fe71d1d50f09348da2dc34941724407a1ffe"
iuc
parents: 2
diff changeset
118 if FILE2SPECIESURL.get(dataset, False):
c29275496061 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit 8533fe71d1d50f09348da2dc34941724407a1ffe"
iuc
parents: 2
diff changeset
119 url_download(FILE2SPECIESURL[dataset], dataset + ".species", workdir)
0
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
120 data_manager_entry = {}
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
121 data_manager_entry['value'] = dataset
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
122 data_manager_entry['name'] = FILE2NAME[dataset]
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
123 data_manager_entry['path'] = dataset + ".species"
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
124 data_manager_json["data_tables"]["dada2_species"] = data_manager_entry
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
125
1
bf7b2c14cabc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit a82e4981dac025c909244acd7127c215bdb519a7"
iuc
parents: 0
diff changeset
126 with open(outjson, 'w') as jf:
bf7b2c14cabc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit a82e4981dac025c909244acd7127c215bdb519a7"
iuc
parents: 0
diff changeset
127 jf.write(json.dumps(data_manager_json, sort_keys=True))
0
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
128
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
129
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
130 if __name__ == '__main__':
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
131 parser = argparse.ArgumentParser(description='Create data manager json.')
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
132 parser.add_argument('--out', action='store', help='JSON filename')
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
133 parser.add_argument('--dataset', action='store', help='Download data set name')
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
134 args = parser.parse_args()
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
135
f57c13f5878b "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_dada2 commit f8b6b6e72914ad6bcca8423dfa03f59bde80992e"
iuc
parents:
diff changeset
136 remote_dataset(args.dataset, args.out)