jdtibochab · jdtibochab · Apr 28, 2025 · Apr 22, 2025 · Apr 22, 2025 · Apr 23, 2025
diff --git a/.vscode/settings.json b/.vscode/settings.json
@@ -0,0 +1,4 @@
+{
+    "python.analysis.typeCheckingMode": "basic",
+    "python.analysis.autoImportCompletions": true
+}
diff --git a/coralme/builder/flat_files.py b/coralme/builder/flat_files.py
@@ -167,7 +167,7 @@ def process_m_model(
 	# cobra.manipulation.delete.prune_unused_metabolites(m_model) # orphan metabolites
 
 	# met_data DataFrame
-	mets_data = mets_data[mets_data['type'].isin(['ADD', 'REPLACE'])]
+	mets_data = mets_data[mets_data['type'].isin(['ADD', 'REPLACE', 'CORRECTION'])]
 	mets_data.columns = ['me_id', 'name', 'formula', 'compartment', 'type']
 	# WARNING: Do we need to rename the index?
 	#mets_data.rename(index = lambda x: x.replace('_DASH_', '__'), inplace = True)

diff --git a/coralme/builder/main.py b/coralme/builder/main.py
@@ -71,7 +71,7 @@ class MEBuilder(object):
 		to update the configuration of the parent class.
 
 	"""
-	def __init__(self, *args, m_model_path = None, genbank_path = None, locus_tag = 'locus_tag', blast_threads = None, **kwargs):
+	def __init__(self, *args, m_model_path = False, genbank_path = False, locus_tag = 'locus_tag', blast_threads = None, **kwargs):
 		"""
 		keyword arguments
 			ME-Model-ID, str
@@ -94,6 +94,11 @@ def __init__(self, *args, m_model_path = None, genbank_path = None, locus_tag =
 			estimate_keffs, default True
 		"""
 
+		self.available_reference_models = {
+				'iJL1678b' : 'locus_tag', # E. coli, gram negative
+				'iJT964'   : 'old_locus_tag', # B. subtilis, gram positive
+				}
+
 		if blast_threads is None:
 			blast_threads = os.cpu_count()-1
 
@@ -265,7 +270,7 @@ def generate_files(self, overwrite = True):
 		if bool(config.get('dev_reference', False)) or bool(config.get('user_reference', False)):
 			logging.warning("Reading reference")
 
-			self.ref = coralme.builder.organism.Organism(config.copy(), is_reference = True)
+			self.ref = coralme.builder.organism.Organism(config.copy(), is_reference = True, available_reference_models = self.available_reference_models)
 			self.ref.get_organism()
 
 			folder = self.org.blast_directory
@@ -1236,8 +1241,9 @@ def update_special_modifications_from_homology(self):
 				ref_cplx_homolog[i] for i in ref_cplxs if i in ref_cplx_homolog
 			]
 			for i in org_cplxs:
-				if v["stoich"]:
-					org_special_trna_subreactions[k]["stoich"] = v["stoich"]
+				# WARNING: stoichiometry of special modifications is in subreactions.txt
+				# if v["stoich"]:
+				# 	org_special_trna_subreactions[k]["stoich"] = v["stoich"]
 				if self._is_base_complex_in_list(i,defined_cplxs):
 					continue
 				defined_cplxs.append(i)

diff --git a/coralme/builder/organism.py b/coralme/builder/organism.py
@@ -5,6 +5,7 @@
 import anyconfig
 import numpy
 import collections
+import pathlib
 
 import Bio
 import cobra
@@ -42,33 +43,47 @@ class Organism(object):
         If True, process as reference organism.
     """
 
-    def __init__(self, config, is_reference):
+    def __init__(self, config, is_reference, available_reference_models = None):
+        self.available_reference_models = available_reference_models
+
         if is_reference:
+            # check values
+            if not bool(config.get('dev_reference', False)) and not bool(config.get('user_reference', False)):
+                config['dev_reference'] = 'iJL1678b'
+                logging.warning('Using \'iJL1678b\' model as reference. See `builder.available_reference_models` for more information.')
+            elif 'dev_reference' in config and 'user_reference' in config:
+                raise ValueError('The keys \'dev_reference\' and \'user_reference\' are mutually exclusive. Please remove one from configuration.')
+            elif config['dev_reference'] is True:
+                config['dev_reference'] = 'iJL1678b'
+                logging.warning('Using \'iJL1678b\' model as reference. See `builder.available_reference_models` for more information.')
+            elif config['dev_reference']:
+                if not config['dev_reference'] in self.available_reference_models:
+                    raise ValueError('The key \'dev_reference\' must be \'iJL1678b\' (E. coli, gram negative) or \'iJT964\' (B. subtilis, gram positive)')
+            elif config['user_reference']:
+                if not pathlib.Path(config['user_reference']).is_dir():
+                    raise ValueError('The key \'user_reference\' must be a directory.')
+            else:
+                return NotImplementedError
+
             # If Organism is reference, set some default values for future processing
-            if bool(config.get('dev_reference', False)) and not bool(config.get('user_reference', False)):
-                # If no reference set, use the default iJL1678b E. coli ME-model files as reference
-                self.id = 'iJL1678b'
-                config["locus_tag"] = "locus_tag"
-            elif not bool(config.get('dev_reference', False)) and bool(config.get('user_reference', False)):
+            if config['dev_reference']:
+                self.id = config['dev_reference']
+                config["locus_tag"] = self.available_reference_models[self.id]
+
+            elif config['user_reference']:
                 # If User set a reference, use manually set values
                 # Set ID
                 self.id = config['user_reference']
                 config = config.copy()
+                # TODO: check if necessary
                 # Update the configuration using the configuration from the User's defined reference
                 for input_file in [config['user_reference'] + "/organism.json", \
                                     config['user_reference'] + "/input.json"]:
                     with open(input_file, 'r') as infile:
                         config.update(anyconfig.load(infile))
-            elif bool(config.get('dev_reference', False)) and bool(config.get('user_reference', False)):
-                # Reference configuration is contradictory. Using default reference instead.
-                logging.warning('The \'dev_reference\' and \'user-reference\' options are mutually exclusive. Using default iJL1678b as reference.')
-                self.id = 'iJL1678b'
-                config["locus_tag"] = "locus_tag"
+
             else:
-                # In any other case, use the default reference instead
-                logging.warning('No reference was set. Using default iJL1678b as reference.')
-                self.id = 'iJL1678b'
-                config["locus_tag"] = "locus_tag"
+                return NotImplementedError
         else:
             # If it is not the reference, use the organism files
             self.id = config['ME-Model-ID']
@@ -104,7 +119,7 @@ def __init__(self, config, is_reference):
     @property
     def directory(self):
         """Working file directory with reconstruction files"""
-        if self.is_reference and self.id == 'iJL1678b':
+        if self.is_reference and self.id in self.available_reference_models:
             try:
                 from importlib.resources import files
             except ImportError:
@@ -188,9 +203,12 @@ def _m_model(self):
         """
         Returns the M-model.
         """
-        if self.id == 'iJL1678b':
-            # If reference organism is iJL1678b, read it from m_model.json
-            model = self.directory + 'm_model.json'
+        if self.is_reference:
+            if self.id in self.available_reference_models:
+                # If reference organism is iJL1678b, read it from m_model.json
+                model = self.directory + 'm_model.json'
+            else:
+                raise ValueError('Reference code is not valid. See builder.available_reference_models for available references.')
         else:
             # Read M-model from the configuration file path
             model = self.config['m-model-path']
@@ -300,9 +318,12 @@ def get_organism(self):
     def get_genbank_contigs(self):
         """ Reads GenBank file as a list of contigs.
         """
-        if self.id == 'iJL1678b':
-            # If default reference iJL1678b, read from genome.gb
-            gb_it = Bio.SeqIO.parse(self.directory + "genome.gb", "gb")
+        if self.is_reference:
+            if self.id in self.available_reference_models:
+                # If default reference iJL1678b, read from genome.gb
+                gb_it = Bio.SeqIO.parse(self.directory + "genome.gb", "gb")
+            else:
+                raise ValueError('Reference code is not valid. See builder.available_reference_models for available references.')
         else:
             # Read from configuration file path
             gb_it = Bio.SeqIO.parse(self.config['genbank-path'], "gb")

diff --git a/coralme/builder/troubleshooting.py b/coralme/builder/troubleshooting.py
@@ -216,7 +216,7 @@ def brute_force_check(me_model, metabolites_to_add, growth_key_and_value,solver=
 	"""
 	if solver in ['gurobi', 'cplex']:
 		me_model.get_solution = me_model.optimize_windows
-		me_model.get_feasibility =me_model.feas_windows(solver = solver)
+		me_model.get_feasibility = me_model.feas_windows(solver = solver)
 	elif solver == "qminos":
 		me_model.get_solution = me_model.optimize
 		me_model.get_feasibility = me_model.feasibility

diff --git a/coralme/core/__init__.py b/coralme/core/__init__.py
@@ -3,3 +3,4 @@
 import coralme.core.parameters
 import coralme.core.processdata
 import coralme.core.reaction
+import coralme.core.optimization