Mercurial > repos > matthias > data_manager_dada2
annotate data_manager/data_manager.py @ 2:b4c303665291 draft
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
author | matthias |
---|---|
date | Fri, 08 Mar 2019 05:38:44 -0500 |
parents | 419037fe1150 |
children | 3a4ee8bf012a |
rev | line source |
---|---|
0
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
1 import argparse |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
2 import json |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
3 import os |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
4 import shutil |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
5 import sys |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
6 import zipfile |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
7 try: |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
8 # For Python 3.0 and later |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
9 from urllib.request import Request, urlopen |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
10 except ImportError: |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
11 # Fall back to Python 2 imports |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
12 from urllib2 import Request, urlopen |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
13 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
14 DEFAULT_TAXLEVELS="Kingdom,Phylum,Class,Order,Family,Genus,Species" |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
15 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
16 FILE2NAME = { |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
17 "silva132":"Silva version 132", |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
18 "silva128":"Silva version 128", |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
19 "rdp16":"RDP trainset 16", |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
20 "rdp14":"RDP trainset 14", |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
21 "gg13.84":"GreenGenes version 13.8", |
2
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
22 "unite8.0_fungi": "UNITE: General Fasta release 8.0 for Fungi", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
23 "unite8.0_fungi_singletons": "UNITE: General Fasta release 8.0 for Fungi including global and 97% singletons", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
24 "unite8.0_euka": "UNITE: General Fasta release 8.0 for all Eukaryotes", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
25 "unite8.0_euka_singletons": "UNITE: General Fasta release 8.0 for all Eukaryotes including global and 97% singletons", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
26 "RefSeq_RDP_2018_05": "NCBI RefSeq 16S rRNA database supplemented by RDP (05/2018)", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
27 "gtdb_2018_11_20": "GTDB: Genome Taxonomy Database (Bacteria & Archaea) (11/2018)", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
28 "hitdb1": "HitDB version 1 (Human InTestinal 16S rRNA)", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
29 "silva132_euk_18S": "Silva version 132 Eukaryotic 18S", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
30 "PR2v4.11.1": "Protist Ribosomal Reference database (PR2) 4.11.1" |
0
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
31 } |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
32 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
33 FILE2TAXURL = { |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
34 "silva132":"https://zenodo.org/record/1172783/files/silva_nr_v132_train_set.fa.gz?download=1", |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
35 "silva128":"https://zenodo.org/record/824551/files/silva_nr_v128_train_set.fa.gz?download=1", |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
36 "rdp16":"https://zenodo.org/record/801828/files/rdp_train_set_16.fa.gz?download=1", |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
37 "rdp14":"https://zenodo.org/record/158955/files/rdp_train_set_14.fa.gz?download=1", |
2
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
38 "unite8.0_fungi": "https://files.plutof.ut.ee/public/orig/EB/0C/EB0CCB3A871B77EA75E472D13926271076904A588D2E1C1EA5AFCF7397D48378.zip", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
39 "unite8.0_fungi_singletons": "https://files.plutof.ut.ee/doi/06/A2/06A2C86256EED64085670EB0C54B7115F6DAC8F311C656A9CB33E386CFABA0D0.zip", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
40 "unite8.0_euka": "https://files.plutof.ut.ee/public/orig/D6/96/D69658E99589D888A207805A744019DBA4EC0F603E67E53732767B3E03A5AA86.zip", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
41 "unite8.0_euka_singletons": "https://files.plutof.ut.ee/doi/C2/20/C22034350E32D6AD7E5D1AF3F8BC487E34DA0BE25602B0E748906005CE6ADA97.zip", |
0
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
42 "gg13.84":"https://zenodo.org/record/158955/files/gg_13_8_train_set_97.fa.gz?download=1", |
2
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
43 "RefSeq_RDP_2018_05": "https://zenodo.org/record/2541239/files/RefSeq-RDP16S_v2_May2018.fa.gz?download=1", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
44 "gtdb_2018_11_20": "https://zenodo.org/record/2541239/files/GTDB_bac-arc_ssu_r86.fa.gz?download=1", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
45 "hitdb1": "https://zenodo.org/record/159205/files/hitdb_v1.00.fa.gz?download=1", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
46 "silva132_euk_18S": "https://zenodo.org/record/1447330/files/silva_132.18s.99_rep_set.dada2.fa.gz?download=1", |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
47 "PR2v4.11.1": "https://github.com/pr2database/pr2database/releases/download/4.11.1/pr2_version_4.11.1_dada2.fasta.gz" |
0
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
48 } |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
49 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
50 FILE2SPECIESURL = { |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
51 "silva132":"https://zenodo.org/record/1172783/files/silva_species_assignment_v132.fa.gz?download=1", |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
52 "silva128":"https://zenodo.org/record/824551/files/silva_species_assignment_v128.fa.gz?download=1", |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
53 "rdp16":"https://zenodo.org/record/801828/files/rdp_species_assignment_16.fa.gz?download=1", |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
54 "rdp14":"https://zenodo.org/record/158955/files/rdp_species_assignment_14.fa.gz?download=1" |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
55 } |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
56 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
57 FILE2TAXLEVELS = { |
2
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
58 "PR2v4.11.1": "Kingdom,Supergroup,Division,Class,Order,Family,Genus,Species" |
0
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
59 } |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
60 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
61 def url_download(url, fname, workdir): |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
62 """ |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
63 download url to workdir/fname |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
64 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
65 return the path to the resulting file |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
66 """ |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
67 file_path = os.path.join(workdir, fname) |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
68 if not os.path.exists(workdir): |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
69 os.makedirs(workdir) |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
70 src = None |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
71 dst = None |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
72 try: |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
73 req = Request(url) |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
74 src = urlopen(req) |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
75 with open(file_path, 'wb') as dst: |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
76 while True: |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
77 chunk = src.read(2**10) |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
78 if chunk: |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
79 dst.write(chunk) |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
80 else: |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
81 break |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
82 finally: |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
83 if src: |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
84 src.close() |
2
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
85 |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
86 #special treatment of UNITE DBs: they are zip files containing two fasta (xyz.fasta and developer/xyz.fasta) |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
87 if fname.startswith("unite"): |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
88 import glob |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
89 import gzip |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
90 import shutil |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
91 import zipfile |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
92 # unzip download |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
93 zip_ref = zipfile.ZipFile(file_path, 'r') |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
94 zip_ref.extractall(workdir) |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
95 zip_ref.close() |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
96 # gzip top level fasta file |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
97 fastas = glob.glob("*fasta") |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
98 if len(fastas) != 1: |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
99 msg = "UNITE download %s contained more than one or no fasta file" |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
100 raise Exception(msg) |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
101 with open(fastas[0], 'rb') as f_in: |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
102 with gzip.open(file_path, 'wb') as f_out: |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
103 shutil.copyfileobj(f_in, f_out) |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
104 |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
105 return fname |
0
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
106 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
107 def main(dataset, outjson): |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
108 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
109 params = json.loads(open(outjson).read()) |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
110 target_directory = params['output_data'][0]['extra_files_path'] |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
111 os.mkdir(target_directory) |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
112 output_path = os.path.abspath(os.path.join(os.getcwd(), 'dada2')) |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
113 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
114 workdir = os.path.join(os.getcwd(), 'dada2') |
2
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
115 path = url_download( FILE2TAXURL[dataset], dataset+".taxonomy", workdir) |
0
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
116 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
117 data_manager_json = {"data_tables":{}} |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
118 data_manager_entry = {} |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
119 data_manager_entry['value'] = dataset |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
120 data_manager_entry['name'] = FILE2NAME[dataset] |
2
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
121 data_manager_entry['path'] = dataset+".taxonomy" |
0
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
122 data_manager_entry['taxlevels'] = FILE2TAXLEVELS.get(dataset, DEFAULT_TAXLEVELS) |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
123 data_manager_json["data_tables"]["dada2_taxonomy"] = data_manager_entry |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
124 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
125 |
2
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
126 if FILE2SPECIESURL.get(dataset, False ): |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
127 path = url_download( FILE2SPECIESURL[dataset], dataset+".species", workdir) |
0
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
128 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
129 data_manager_entry = {} |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
130 data_manager_entry['value'] = dataset |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
131 data_manager_entry['name'] = FILE2NAME[dataset] |
2
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
132 data_manager_entry['path'] = dataset+".species" |
0
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
133 data_manager_json["data_tables"]["dada2_species"] = data_manager_entry |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
134 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
135 for filename in os.listdir(workdir): |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
136 shutil.move(os.path.join(output_path, filename), target_directory) |
2
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
137 |
b4c303665291
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit eec95ccc2189355061112ea2785b82f13a0fa077-dirty
matthias
parents:
0
diff
changeset
|
138 sys.stderr.write("JSON %s" %json.dumps(data_manager_json)) |
0
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
139 file(outjson, 'w').write(json.dumps(data_manager_json)) |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
140 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
141 if __name__ == '__main__': |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
142 parser = argparse.ArgumentParser(description='Create data manager json.') |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
143 parser.add_argument('--out', action='store', help='JSON filename') |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
144 parser.add_argument('--dataset', action='store', help='Download data set name') |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
145 args = parser.parse_args() |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
146 |
419037fe1150
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/data_managers/data_manager_dada2 commit 42eb67646e47bef13eed672ff6b9d06b1d82ae3d-dirty
matthias
parents:
diff
changeset
|
147 main(args.dataset, args.out) |