|
542
|
1 """
|
|
|
2 Apply RAS-based scaling to reaction bounds and optionally save updated models.
|
|
|
3
|
|
|
4 Workflow:
|
|
|
5 - Read one or more RAS matrices (patients/samples x reactions)
|
|
|
6 - Normalize and merge them, optionally adding class suffixes to sample IDs
|
|
|
7 - Build a COBRA model from a tabular CSV
|
|
|
8 - Run FVA to initialize bounds, then scale per-sample based on RAS values
|
|
|
9 - Save bounds per sample and optionally export updated models in chosen formats
|
|
|
10 """
|
|
|
11 import argparse
|
|
|
12 from typing import Optional, Dict, Set, List, Tuple, Union
|
|
|
13 import os
|
|
|
14 import numpy as np
|
|
|
15 import pandas as pd
|
|
|
16 import cobra
|
|
|
17 from cobra import Model
|
|
|
18 import sys
|
|
|
19 from joblib import Parallel, delayed, cpu_count
|
|
|
20
|
|
|
21 try:
|
|
|
22 from .utils import general_utils as utils
|
|
|
23 from .utils import model_utils as modelUtils
|
|
|
24 except:
|
|
|
25 import utils.general_utils as utils
|
|
|
26 import utils.model_utils as modelUtils
|
|
|
27
|
|
|
28 ################################# process args ###############################
|
|
|
29 def process_args(args :List[str] = None) -> argparse.Namespace:
|
|
|
30 """
|
|
|
31 Processes command-line arguments.
|
|
|
32
|
|
|
33 Args:
|
|
|
34 args (list): List of command-line arguments.
|
|
|
35
|
|
|
36 Returns:
|
|
|
37 Namespace: An object containing parsed arguments.
|
|
|
38 """
|
|
|
39 parser = argparse.ArgumentParser(usage = '%(prog)s [options]',
|
|
|
40 description = 'process some value\'s')
|
|
|
41
|
|
|
42
|
|
|
43 parser.add_argument("-mo", "--model_upload", type = str,
|
|
|
44 help = "path to input file with custom rules, if provided")
|
|
|
45
|
|
|
46 parser.add_argument('-ol', '--out_log',
|
|
|
47 help = "Output log")
|
|
|
48
|
|
|
49 parser.add_argument('-td', '--tool_dir',
|
|
|
50 type = str,
|
|
|
51 default = os.path.dirname(os.path.abspath(__file__)),
|
|
|
52 help = 'your tool directory (default: auto-detected package location)')
|
|
|
53
|
|
|
54 parser.add_argument('-ir', '--input_ras',
|
|
|
55 type=str,
|
|
|
56 required = False,
|
|
|
57 help = 'input ras')
|
|
|
58
|
|
|
59 parser.add_argument('-rn', '--name',
|
|
|
60 type=str,
|
|
|
61 help = 'ras class names')
|
|
|
62
|
|
|
63 parser.add_argument('-cc', '--cell_class',
|
|
|
64 type = str,
|
|
|
65 help = 'output of cell class')
|
|
|
66 parser.add_argument(
|
|
|
67 '-idop', '--output_path',
|
|
|
68 type = str,
|
|
|
69 default='ras_to_bounds/',
|
|
|
70 help = 'output path for maps')
|
|
|
71
|
|
|
72 parser.add_argument('-sm', '--save_models',
|
|
|
73 type=utils.Bool("save_models"),
|
|
|
74 default=False,
|
|
|
75 help = 'whether to save models with applied bounds')
|
|
|
76
|
|
|
77 parser.add_argument('-smp', '--save_models_path',
|
|
|
78 type = str,
|
|
|
79 default='saved_models/',
|
|
|
80 help = 'output path for saved models')
|
|
|
81
|
|
|
82 parser.add_argument('-smf', '--save_models_format',
|
|
|
83 type = str,
|
|
|
84 default='csv',
|
|
|
85 help = 'format for saved models (csv, xml, json, mat, yaml, tabular)')
|
|
|
86
|
|
|
87
|
|
|
88 ARGS = parser.parse_args(args)
|
|
|
89 return ARGS
|
|
|
90
|
|
|
91 ########################### warning ###########################################
|
|
|
92 def warning(s :str) -> None:
|
|
|
93 """
|
|
|
94 Log a warning message to an output log file and print it to the console.
|
|
|
95
|
|
|
96 Args:
|
|
|
97 s (str): The warning message to be logged and printed.
|
|
|
98
|
|
|
99 Returns:
|
|
|
100 None
|
|
|
101 """
|
|
|
102 if ARGS.out_log:
|
|
|
103 with open(ARGS.out_log, 'a') as log:
|
|
|
104 log.write(s + "\n\n")
|
|
|
105 print(s)
|
|
|
106
|
|
|
107 ############################ dataset input ####################################
|
|
|
108 def read_dataset(data :str, name :str) -> pd.DataFrame:
|
|
|
109 """
|
|
|
110 Read a dataset from a CSV file and return it as a pandas DataFrame.
|
|
|
111
|
|
|
112 Args:
|
|
|
113 data (str): Path to the CSV file containing the dataset.
|
|
|
114 name (str): Name of the dataset, used in error messages.
|
|
|
115
|
|
|
116 Returns:
|
|
|
117 pandas.DataFrame: DataFrame containing the dataset.
|
|
|
118
|
|
|
119 Raises:
|
|
|
120 pd.errors.EmptyDataError: If the CSV file is empty.
|
|
|
121 sys.exit: If the CSV file has the wrong format, the execution is aborted.
|
|
|
122 """
|
|
|
123 try:
|
|
|
124 dataset = pd.read_csv(data, sep = '\t', header = 0, engine='python')
|
|
|
125 except pd.errors.EmptyDataError:
|
|
|
126 sys.exit('Execution aborted: wrong format of ' + name + '\n')
|
|
|
127 if len(dataset.columns) < 2:
|
|
|
128 sys.exit('Execution aborted: wrong format of ' + name + '\n')
|
|
|
129 return dataset
|
|
|
130
|
|
|
131
|
|
|
132 def apply_ras_bounds(bounds, ras_row):
|
|
|
133 """
|
|
|
134 Adjust the bounds of reactions in the model based on RAS values.
|
|
|
135
|
|
|
136 Args:
|
|
|
137 bounds (pd.DataFrame): Model bounds.
|
|
|
138 ras_row (pd.Series): A row from a RAS DataFrame containing scaling factors for reaction bounds.
|
|
|
139 Returns:
|
|
|
140 new_bounds (pd.DataFrame): integrated bounds.
|
|
|
141 """
|
|
|
142 new_bounds = bounds.copy()
|
|
|
143 for reaction in ras_row.index:
|
|
|
144 scaling_factor = ras_row[reaction]
|
|
|
145 if not np.isnan(scaling_factor):
|
|
|
146 lower_bound=bounds.loc[reaction, "lower_bound"]
|
|
|
147 upper_bound=bounds.loc[reaction, "upper_bound"]
|
|
|
148 valMax=float((upper_bound)*scaling_factor)
|
|
|
149 valMin=float((lower_bound)*scaling_factor)
|
|
|
150 if upper_bound!=0 and lower_bound==0:
|
|
|
151 new_bounds.loc[reaction, "upper_bound"] = valMax
|
|
|
152 if upper_bound==0 and lower_bound!=0:
|
|
|
153 new_bounds.loc[reaction, "lower_bound"] = valMin
|
|
|
154 if upper_bound!=0 and lower_bound!=0:
|
|
|
155 new_bounds.loc[reaction, "lower_bound"] = valMin
|
|
|
156 new_bounds.loc[reaction, "upper_bound"] = valMax
|
|
|
157 return new_bounds
|
|
|
158
|
|
|
159
|
|
|
160 def save_model(model, filename, output_folder, file_format='csv'):
|
|
|
161 """
|
|
|
162 Save a COBRA model to file in the specified format.
|
|
|
163
|
|
|
164 Args:
|
|
|
165 model (cobra.Model): The model to save.
|
|
|
166 filename (str): Base filename (without extension).
|
|
|
167 output_folder (str): Output directory.
|
|
|
168 file_format (str): File format ('xml', 'json', 'mat', 'yaml', 'tabular', 'csv').
|
|
|
169
|
|
|
170 Returns:
|
|
|
171 None
|
|
|
172 """
|
|
|
173 if not os.path.exists(output_folder):
|
|
|
174 os.makedirs(output_folder)
|
|
|
175
|
|
|
176 try:
|
|
|
177 if file_format == 'tabular' or file_format == 'csv':
|
|
|
178 # Special handling for tabular format using utils functions
|
|
|
179 filepath = os.path.join(output_folder, f"{filename}.csv")
|
|
|
180
|
|
|
181 # Use unified function for tabular export
|
|
|
182 merged = modelUtils.export_model_to_tabular(
|
|
|
183 model=model,
|
|
|
184 output_path=filepath,
|
|
|
185 include_objective=True
|
|
|
186 )
|
|
|
187
|
|
|
188 else:
|
|
|
189 # Standard COBRA formats
|
|
|
190 filepath = os.path.join(output_folder, f"{filename}.{file_format}")
|
|
|
191
|
|
|
192 if file_format == 'xml':
|
|
|
193 cobra.io.write_sbml_model(model, filepath)
|
|
|
194 elif file_format == 'json':
|
|
|
195 cobra.io.save_json_model(model, filepath)
|
|
|
196 elif file_format == 'mat':
|
|
|
197 cobra.io.save_matlab_model(model, filepath)
|
|
|
198 elif file_format == 'yaml':
|
|
|
199 cobra.io.save_yaml_model(model, filepath)
|
|
|
200 else:
|
|
|
201 raise ValueError(f"Unsupported format: {file_format}")
|
|
|
202
|
|
|
203 print(f"Model saved: {filepath}")
|
|
|
204
|
|
|
205 except Exception as e:
|
|
|
206 warning(f"Error saving model {filename}: {str(e)}")
|
|
|
207
|
|
|
208 def apply_bounds_to_model(model, bounds):
|
|
|
209 """
|
|
|
210 Apply bounds from a DataFrame to a COBRA model.
|
|
|
211
|
|
|
212 Args:
|
|
|
213 model (cobra.Model): The metabolic model to modify.
|
|
|
214 bounds (pd.DataFrame): DataFrame with reaction bounds.
|
|
|
215
|
|
|
216 Returns:
|
|
|
217 cobra.Model: Modified model with new bounds.
|
|
|
218 """
|
|
|
219 model_copy = model.copy()
|
|
|
220 for reaction_id in bounds.index:
|
|
|
221 try:
|
|
|
222 reaction = model_copy.reactions.get_by_id(reaction_id)
|
|
|
223 reaction.lower_bound = bounds.loc[reaction_id, "lower_bound"]
|
|
|
224 reaction.upper_bound = bounds.loc[reaction_id, "upper_bound"]
|
|
|
225 except KeyError:
|
|
|
226 # Reaction not found in model, skip
|
|
|
227 continue
|
|
|
228 return model_copy
|
|
|
229
|
|
|
230 def process_ras_cell(cellName, ras_row, model, rxns_ids, output_folder, save_models=False, save_models_path='saved_models/', save_models_format='csv'):
|
|
|
231 """
|
|
|
232 Process a single RAS cell, apply bounds, and save the bounds to a CSV file.
|
|
|
233
|
|
|
234 Args:
|
|
|
235 cellName (str): The name of the RAS cell (used for naming the output file).
|
|
|
236 ras_row (pd.Series): A row from a RAS DataFrame containing scaling factors for reaction bounds.
|
|
|
237 model (cobra.Model): The metabolic model to be modified.
|
|
|
238 rxns_ids (list of str): List of reaction IDs to which the scaling factors will be applied.
|
|
|
239 output_folder (str): Folder path where the output CSV file will be saved.
|
|
|
240 save_models (bool): Whether to save models with applied bounds.
|
|
|
241 save_models_path (str): Path where to save models.
|
|
|
242 save_models_format (str): Format for saved models.
|
|
|
243
|
|
|
244 Returns:
|
|
|
245 None
|
|
|
246 """
|
|
|
247 bounds = pd.DataFrame([(rxn.lower_bound, rxn.upper_bound) for rxn in model.reactions], index=rxns_ids, columns=["lower_bound", "upper_bound"])
|
|
|
248 new_bounds = apply_ras_bounds(bounds, ras_row)
|
|
|
249 new_bounds.to_csv(output_folder + cellName + ".csv", sep='\t', index=True)
|
|
|
250
|
|
|
251 # Save model if requested
|
|
|
252 if save_models:
|
|
|
253 modified_model = apply_bounds_to_model(model, new_bounds)
|
|
|
254 save_model(modified_model, cellName, save_models_path, save_models_format)
|
|
|
255
|
|
|
256 return
|
|
|
257
|
|
|
258 def generate_bounds_model(model: cobra.Model, ras=None, output_folder='output/', save_models=False, save_models_path='saved_models/', save_models_format='csv') -> pd.DataFrame:
|
|
|
259 """
|
|
|
260 Generate reaction bounds for a metabolic model based on medium conditions and optional RAS adjustments.
|
|
|
261
|
|
|
262 Args:
|
|
|
263 model (cobra.Model): The metabolic model for which bounds will be generated.
|
|
|
264 ras (pd.DataFrame, optional): RAS pandas dataframe. Defaults to None.
|
|
|
265 output_folder (str, optional): Folder path where output CSV files will be saved. Defaults to 'output/'.
|
|
|
266 save_models (bool): Whether to save models with applied bounds.
|
|
|
267 save_models_path (str): Path where to save models.
|
|
|
268 save_models_format (str): Format for saved models.
|
|
|
269
|
|
|
270 Returns:
|
|
|
271 pd.DataFrame: DataFrame containing the bounds of reactions in the model.
|
|
|
272 """
|
|
|
273 rxns_ids = [rxn.id for rxn in model.reactions]
|
|
|
274
|
|
|
275 # Perform Flux Variability Analysis (FVA) on this medium
|
|
|
276 df_FVA = cobra.flux_analysis.flux_variability_analysis(model, fraction_of_optimum=0, processes=1).round(8)
|
|
|
277
|
|
|
278 # Set FVA bounds
|
|
|
279 for reaction in rxns_ids:
|
|
|
280 model.reactions.get_by_id(reaction).lower_bound = float(df_FVA.loc[reaction, "minimum"])
|
|
|
281 model.reactions.get_by_id(reaction).upper_bound = float(df_FVA.loc[reaction, "maximum"])
|
|
|
282
|
|
|
283 if ras is not None:
|
|
|
284 Parallel(n_jobs=cpu_count())(delayed(process_ras_cell)(
|
|
|
285 cellName, ras_row, model, rxns_ids, output_folder,
|
|
|
286 save_models, save_models_path, save_models_format
|
|
|
287 ) for cellName, ras_row in ras.iterrows())
|
|
|
288 else:
|
|
|
289 raise ValueError("RAS DataFrame is None. Cannot generate bounds without RAS data.")
|
|
|
290 return
|
|
|
291
|
|
|
292 ############################# main ###########################################
|
|
|
293 def main(args:List[str] = None) -> None:
|
|
|
294 """
|
|
|
295 Initialize and execute RAS-to-bounds pipeline based on the frontend input arguments.
|
|
|
296
|
|
|
297 Returns:
|
|
|
298 None
|
|
|
299 """
|
|
|
300 if not os.path.exists('ras_to_bounds'):
|
|
|
301 os.makedirs('ras_to_bounds')
|
|
|
302
|
|
|
303 global ARGS
|
|
|
304 ARGS = process_args(args)
|
|
|
305
|
|
|
306
|
|
|
307 ras_file_list = ARGS.input_ras.split(",")
|
|
|
308 ras_file_names = ARGS.name.split(",")
|
|
|
309 if len(ras_file_names) != len(set(ras_file_names)):
|
|
|
310 error_message = "Duplicated file names in the uploaded RAS matrices."
|
|
|
311 warning(error_message)
|
|
|
312 raise ValueError(error_message)
|
|
|
313
|
|
|
314 ras_class_names = []
|
|
|
315 for file in ras_file_names:
|
|
|
316 ras_class_names.append(file.rsplit(".", 1)[0])
|
|
|
317 ras_list = []
|
|
|
318 class_assignments = pd.DataFrame(columns=["Patient_ID", "Class"])
|
|
|
319 for ras_matrix, ras_class_name in zip(ras_file_list, ras_class_names):
|
|
|
320 ras = read_dataset(ras_matrix, "ras dataset")
|
|
|
321 ras.replace("None", None, inplace=True)
|
|
|
322 ras.set_index("Reactions", drop=True, inplace=True)
|
|
|
323 ras = ras.T
|
|
|
324 ras = ras.astype(float)
|
|
|
325 if(len(ras_file_list)>1):
|
|
|
326 # Append class name to patient id (DataFrame index)
|
|
|
327 ras.index = [f"{idx}_{ras_class_name}" for idx in ras.index]
|
|
|
328 else:
|
|
|
329 ras.index = [f"{idx}" for idx in ras.index]
|
|
|
330 ras_list.append(ras)
|
|
|
331 for patient_id in ras.index:
|
|
|
332 class_assignments.loc[class_assignments.shape[0]] = [patient_id, ras_class_name]
|
|
|
333
|
|
|
334
|
|
|
335 # Concatenate all RAS DataFrames into a single DataFrame
|
|
|
336 ras_combined = pd.concat(ras_list, axis=0)
|
|
|
337 # Normalize RAS values column-wise by max RAS
|
|
|
338 ras_combined = ras_combined.div(ras_combined.max(axis=0))
|
|
|
339 ras_combined.dropna(axis=1, how='all', inplace=True)
|
|
|
340
|
|
|
341 model = modelUtils.build_cobra_model_from_csv(ARGS.model_upload)
|
|
|
342
|
|
|
343 validation = modelUtils.validate_model(model)
|
|
|
344
|
|
|
345 print("\n=== MODEL VALIDATION ===")
|
|
|
346 for key, value in validation.items():
|
|
|
347 print(f"{key}: {value}")
|
|
|
348
|
|
|
349
|
|
|
350 generate_bounds_model(model, ras=ras_combined, output_folder=ARGS.output_path,
|
|
|
351 save_models=ARGS.save_models, save_models_path=ARGS.save_models_path,
|
|
|
352 save_models_format=ARGS.save_models_format)
|
|
|
353 class_assignments.to_csv(ARGS.cell_class, sep='\t', index=False)
|
|
|
354
|
|
|
355
|
|
|
356 return
|
|
|
357
|
|
|
358 ##############################################################################
|
|
|
359 if __name__ == "__main__":
|
|
539
|
360 main() |
