annotate COBRAxy/src/exportMetabolicModel.py @ 542:fcdbc81feb45 draft

Uploaded
author francesco_lapi
date Sun, 26 Oct 2025 19:27:41 +0000
parents 7d5b35c715e8
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
542
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
1 """
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
2 Convert a tabular (CSV/TSV/Tabular) description of a COBRA model into a COBRA file.
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
3
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
4 Supported output formats: SBML, JSON, MATLAB (.mat), YAML.
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
5 The script logs to a user-provided file for easier debugging in Galaxy.
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
6 """
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
7
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
8 import os
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
9 import cobra
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
10 import argparse
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
11 from typing import List
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
12 import logging
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
13
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
14 try:
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
15 from .utils import model_utils as modelUtils
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
16 from .utils import general_utils as utils
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
17 except:
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
18 import utils.model_utils as modelUtils
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
19 import utils.general_utils as utils
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
20
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
21 ARGS : argparse.Namespace
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
22 def process_args(args: List[str] = None) -> argparse.Namespace:
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
23 """
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
24 Parse command-line arguments for the CSV-to-COBRA conversion tool.
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
25
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
26 Returns:
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
27 argparse.Namespace: Parsed arguments.
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
28 """
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
29 parser = argparse.ArgumentParser(
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
30 usage="%(prog)s [options]",
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
31 description="Convert a tabular/CSV file to a COBRA model"
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
32 )
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
33
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
34
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
35 parser.add_argument("--out_log", type=str, required=True,
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
36 help="Output log file")
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
37
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
38
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
39 parser.add_argument("--input", type=str, required=True,
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
40 help="Input tabular file (CSV/TSV)")
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
41
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
42
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
43 parser.add_argument("--format", type=str, required=True, choices=["sbml", "json", "mat", "yaml"],
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
44 help="Model format (SBML, JSON, MATLAB, YAML)")
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
45
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
46
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
47 parser.add_argument("--output", type=str, required=True,
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
48 help="Output model file path")
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
49
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
50 parser.add_argument("--tool_dir", type=str, default=os.path.dirname(os.path.abspath(__file__)),
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
51 help="Tool directory (default: auto-detected package location)")
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
52
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
53
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
54 return parser.parse_args(args)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
55
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
56
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
57 ###############################- ENTRY POINT -################################
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
58
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
59 def main(args: List[str] = None) -> None:
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
60 """
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
61 Entry point: parse arguments, build the COBRA model from a CSV/TSV file,
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
62 and save it in the requested format.
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
63
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
64 Returns:
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
65 None
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
66 """
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
67 global ARGS
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
68 ARGS = process_args(args)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
69
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
70 # configure logging to the requested log file (overwrite each run)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
71 logging.basicConfig(filename=ARGS.out_log,
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
72 level=logging.DEBUG,
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
73 format='%(asctime)s %(levelname)s: %(message)s',
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
74 filemode='w')
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
75
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
76 logging.info('Starting fromCSVtoCOBRA tool')
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
77 logging.debug('Args: input=%s format=%s output=%s tool_dir=%s', ARGS.input, ARGS.format, ARGS.output, ARGS.tool_dir)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
78
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
79 try:
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
80 # Basic sanity checks
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
81 if not os.path.exists(ARGS.input):
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
82 logging.error('Input file not found: %s', ARGS.input)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
83
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
84 out_dir = os.path.dirname(os.path.abspath(ARGS.output))
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
85
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
86 if out_dir and not os.path.isdir(out_dir):
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
87 try:
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
88 os.makedirs(out_dir, exist_ok=True)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
89 logging.info('Created missing output directory: %s', out_dir)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
90 except Exception as e:
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
91 logging.exception('Cannot create output directory: %s', out_dir)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
92
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
93 model = modelUtils.build_cobra_model_from_csv(ARGS.input)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
94
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
95
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
96 logging.info('Created model with name: %s (ID: %s)', model.name, model.id)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
97
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
98 # Save model in requested format - Galaxy handles the filename
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
99 if ARGS.format == "sbml":
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
100 cobra.io.write_sbml_model(model, ARGS.output)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
101 elif ARGS.format == "json":
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
102 cobra.io.save_json_model(model, ARGS.output)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
103 elif ARGS.format == "mat":
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
104 cobra.io.save_matlab_model(model, ARGS.output)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
105 elif ARGS.format == "yaml":
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
106 cobra.io.save_yaml_model(model, ARGS.output)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
107 else:
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
108 logging.error('Unknown format requested: %s', ARGS.format)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
109 raise ValueError(f"Unknown format: {ARGS.format}")
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
110
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
111
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
112 logging.info('Model successfully written to %s (format=%s)', ARGS.output, ARGS.format)
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
113 print(f"Model created successfully in {ARGS.format.upper()} format")
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
114
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
115 except Exception as e:
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
116 # Log full traceback to the out_log so Galaxy users/admins can see what happened
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
117 logging.exception('Unhandled exception in fromCSVtoCOBRA')
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
118 print(f"ERROR: {str(e)}")
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
119 raise
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
120
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
121
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
122 if __name__ == '__main__':
fcdbc81feb45 Uploaded
francesco_lapi
parents: 540
diff changeset
123 main()