comparison data_manager/extract.py @ 5:5459ec34767a draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/data_managers/data_manager_funannotate commit 8341270dd36185ebf59d15282bc79f1215e936a4
author iuc
date Fri, 07 Mar 2025 22:33:29 +0000
parents
children
comparison
equal deleted inserted replaced
4:1c81a82b5b8e 5:5459ec34767a
1 import json
2 import os
3 import sys
4
5 fun_db = sys.argv[1]
6 fun_db_value = sys.argv[2]
7 dmjson = sys.argv[3]
8
9 content = []
10 # get options for parameter --busco_db
11 # which are just the subfolders of the db dir (minus outgroups/ and trained_species/)
12 # https://github.com/nextgenusfs/funannotate/blob/8cc40728fee61566fdf736c1f2292e14cc117660/funannotate/predict.py#L319
13 for d in os.scandir(fun_db):
14 if not d.is_dir():
15 continue
16 if d.name in ['outgroups', 'trained_species']:
17 continue
18 if not os.path.exists(os.path.join(d, "dataset.cfg")):
19 continue
20 name = d.name.replace("_", " ").capitalize()
21 content.append({'value': d.name, 'name': name, 'select': 'busco_db', 'db_value': fun_db_value})
22
23 # --busco_seed_species
24 # trained_species
25 for d in os.scandir(os.path.join(fun_db, "trained_species")):
26 if not d.is_dir():
27 continue
28 if not os.path.exists(os.path.join(d, "info.json")):
29 continue
30 name = d.name.replace("_", " ").capitalize()
31 content.append({'value': d.name, 'name': name, 'select': 'trained_species', 'db_value': fun_db_value})
32
33 # --busco_seed_species
34 # outgroups
35 for f in os.scandir(os.path.join(fun_db, "outgroups")):
36 if f.is_dir():
37 continue
38 if not f.name.endswith("_buscos.fa"):
39 continue
40 value = f.name[:-10]
41 name = ' - '.join([x.replace("_", " ").capitalize() for x in value.split('.')])
42 content.append({'value': value, 'name': name, 'select': 'outgroup', 'db_value': fun_db_value})
43
44 with open(dmjson, "w") as fh:
45 json.dump({"data_tables": {"funannotate_options": content}}, fh)
46
47 print(f'{len([c for c in content if c["select"]=="busco_db"])} x busco_db\n')
48 print(f'{len([c for c in content if c["select"]=="trained_species"])} x trained_species\n')
49 print(f'{len([c for c in content if c["select"]=="outgroup"])} x outgroup\n')