def load_gene_synonym(session, gene_db, synonym, data_source_id):
"""Load the synonym for this gene from the given genome."""
data_source_id = get_or_create_data_source(session, data_source_id)
synonym_db, _ = get_or_create(session, Synonym,
type='gene',
ome_id=gene_db.id,
synonym=synonym,
data_source_id=data_source_id)
return synonym_db.id
def load_gene_synonym(session, gene_db, synonym, data_source_id):
"""Load the synonym for this gene from the given genome."""
data_source_id = get_or_create_data_source(session, data_source_id)
synonym_db, _ = get_or_create(session,
Synonym,
type='gene',
ome_id=gene_db.id,
synonym=synonym,
data_source_id=data_source_id)
return synonym_db.id
def __init__(self, name, data_source_id=None, group_name=None, attributes=None):
session = Session()
if data_source_id is None:
data_source, exists = get_or_create(session, DataSource, cobra_id='-1',
name='generic', url_prefix='')
data_source_id = data_source.id
session.close()
self.name = name
self.data_source_id = data_source_id
self.group_name = group_name
self.attributes = attributes
def load_reactions(session, model_db_id, model, old_reaction_ids,
comp_comp_db_ids, final_metabolite_ids):
"""Load the reactions and stoichiometries into the model.
TODO if the reaction is already loaded, we need to check the stoichometry
has. If that doesn't match, then add a new reaction with an incremented ID
(e.g. ACALD_1)
Arguments
---------
session: An SQLAlchemy session.
model_db_id: The database ID for the model.
model: The COBRApy model.
old_reaction_ids: A dictionary where keys are new IDs and values are old IDs
for reactions.
comp_comp_db_ids: A dictionary where keys are the original compartmentalized
metabolite ids and the values are the database IDs for the compartmentalized
components.
final_metabolite_ids: A new dictionary where keys are original
compartmentalized metabolite IDs from the model and values are the new
compartmentalized metabolite IDs.
Returns
-------
A dictionary with keys for reaction BiGG IDs in the model and values for the
associated ModelReaction.id in the database.
"""
# only grab this once
data_source_id = get_or_create_data_source(session, 'old_bigg_id')
# get reaction hash_prefs
hash_prefs = load_tsv(settings.reaction_hash_prefs)
def _check_hash_prefs(a_hash, is_pseudoreaction):
"""Return the preferred BiGG ID for a_hash, or None."""
for row in hash_prefs:
marked_pseudo = len(row) > 2 and row[2] == 'pseudoreaction'
if row[0] == a_hash and marked_pseudo == is_pseudoreaction:
return row[1]
return None
# Generate reaction hashes, and find reactions in the same model in opposite
# directions.
reaction_hashes = {r.id: parse.hash_reaction(r, final_metabolite_ids)
for r in model.reactions}
reverse_reaction_hashes = {r.id: parse.hash_reaction(r, final_metabolite_ids, reverse=True)
for r in model.reactions}
reverse_reaction_hashes_rev = {v: k for k, v in six.iteritems(reverse_reaction_hashes)}
reactions_not_to_reverse = set()
for r_id, h in six.iteritems(reaction_hashes):
if h in reverse_reaction_hashes_rev:
reactions_not_to_reverse.add(r_id)
reactions_not_to_reverse.add(reverse_reaction_hashes_rev[h])
model_db_rxn_ids = {}
for reaction in model.reactions:
# Drop duplicates label
reaction_id = parse.remove_duplicate_tag(reaction.id)
# Get the reaction
reaction_db = (session
.query(Reaction)
.filter(Reaction.bigg_id == reaction_id)
.first())
# check for pseudoreaction
is_pseudoreaction = check_pseudoreaction(reaction_id)
# calculate the hash
reaction_hash = reaction_hashes[reaction.id]
hash_db = (session
.query(Reaction)
.filter(Reaction.reaction_hash == reaction_hash)
.filter(Reaction.pseudoreaction == is_pseudoreaction)
.first())
# If there wasn't a match for the forward hash, also check the reverse
# hash. Do not check reverse hash for reactions with both directions
# defined in the same model (e.g. SUCDi and FRD7).
if not hash_db and reaction.id not in reactions_not_to_reverse:
reverse_hash_db = (session
.query(Reaction)
.filter(Reaction.reaction_hash == reverse_reaction_hashes[reaction.id])
.filter(Reaction.pseudoreaction == is_pseudoreaction)
.first())
else:
reverse_hash_db = None
# bigg_id match hash match b==h pseudoreaction example function
# n n n first GAPD _new_reaction (1)
# n n y first EX_glc_e _new_reaction (1)
# y n n incorrect GAPD _new_reaction & increment (2)
# y n y incorrect EX_glc_e _new_reaction & increment (2)
# n y n GAPDH after GAPD reaction = hash_reaction (3a)
# n y y EX_glc__e after EX_glc_e reaction = hash_reaction (3a)
# y y n n ? reaction = hash_reaction (3a)
# y y n y ? reaction = hash_reaction (3a)
# y y y n second GAPD reaction = bigg_reaction (3b)
# y y y y second EX_glc_e reaction = bigg_reaction (3b)
# NOTE: only check pseudoreaction hash against other pseudoreactions
# 4a and 4b are 3a and 3b with a reversed reaction
def _find_new_incremented_id(session, original_id):
"""Look for a reaction bigg_id that is not already taken."""
new_id = increment_id(original_id)
while True:
# Check for existing and deprecated reaction ids
if (session.query(Reaction).filter(Reaction.bigg_id == new_id).first() is None and
not _is_deprecated_reaction_id(session, new_id)):
return new_id
new_id = increment_id(new_id)
# Check for a preferred ID in the preferences, based on the forward
# hash. Don't check the reverse hash in preferences.
preferred_id = _check_hash_prefs(reaction_hash, is_pseudoreaction)
# no reversed by default
is_reversed = False
is_new = False
# (0) If there is a preferred ID, make that the new ID, and increment any old IDs
if preferred_id is not None:
# if the reaction already matches, just continue
if hash_db is not None and hash_db.bigg_id == preferred_id:
reaction_db = hash_db
# otherwise, make the new reaction
else:
# if existing reactions match the preferred reaction find a new,
# incremented id for the existing match
preferred_id_db = session.query(Reaction).filter(Reaction.bigg_id == preferred_id).first()
if preferred_id_db is not None:
new_id = _find_new_incremented_id(session, preferred_id)
logging.warning('Incrementing database reaction {} to {} and prefering {} (from model {}) based on hash preferences'
.format(preferred_id, new_id, preferred_id, model.id))
preferred_id_db.bigg_id = new_id
session.commit()
# make a new reaction for the preferred_id
reaction_db = _new_reaction(session, reaction, preferred_id,
reaction_hash, model_db_id, model,
is_pseudoreaction, comp_comp_db_ids)
is_new = True
# (1) no bigg_id matches, no stoichiometry match or pseudoreaction, then
# make a new reaction
elif reaction_db is None and hash_db is None and reverse_hash_db is None:
# check that the id is not deprecated
if _is_deprecated_reaction_id(session, reaction.id):
logging.error(('Keeping bigg_id {} (hash {} - from model {}) '
'even though it is on the deprecated ID list. '
'You should add it to reaction-hash-prefs.txt')
.format(reaction_id, reaction_hash, model.id))
reaction_db = _new_reaction(session, reaction, reaction_id,
reaction_hash, model_db_id, model,
is_pseudoreaction, comp_comp_db_ids)
is_new = True
# (2) bigg_id matches, but not the hash, then increment the BIGG_ID
elif reaction_db is not None and hash_db is None and reverse_hash_db is None:
# loop until we find a non-matching find non-matching ID
new_id = _find_new_incremented_id(session, reaction.id)
logging.warning('Incrementing bigg_id {} to {} (from model {}) based on conflicting reaction hash'
.format(reaction_id, new_id, model.id))
reaction_db = _new_reaction(session, reaction, new_id,
reaction_hash, model_db_id, model,
is_pseudoreaction, comp_comp_db_ids)
is_new = True
# (3) but found a stoichiometry match, then use the hash reaction match.
elif hash_db is not None:
# WARNING TODO this requires that loaded metabolites always match on
# bigg_id, which should be the case.
# (3a)
if reaction_db is None or reaction_db.id != hash_db.id:
reaction_db = hash_db
# (3b) BIGG ID matches a reaction with the same hash, then just continue
else:
pass
# (4) but found a stoichiometry match, then use the hash reaction match.
elif reverse_hash_db is not None:
# WARNING TODO this requires that loaded metabolites always match on
# bigg_id, which should be the case.
# Remember to switch upper and lower bounds
is_reversed = True
logging.info('Matched {} to {} based on reverse hash'
.format(reaction_id, reverse_hash_db.bigg_id))
# (4a)
if reaction_db is None or reaction_db.id != reverse_hash_db.id:
reaction_db = reverse_hash_db
# (4b) BIGG ID matches a reaction with the same hash, then just continue
else:
pass
else:
raise Exception('Should not get here')
# If the reaction is not new, consider improving the descriptive name
if not is_new:
new_name = scrub_name(check_none(getattr(reaction, 'name', None)))
improve_name(session, reaction_db, new_name)
# Add reaction to deprecated ID list if necessary
if reaction_db.bigg_id != reaction_id:
get_or_create(session, DeprecatedID, deprecated_id=reaction_id,
type='reaction', ome_id=reaction_db.id)
# If the reaction is reversed, then switch upper and lower bound
lower_bound = -reaction.upper_bound if is_reversed else reaction.lower_bound
upper_bound = -reaction.lower_bound if is_reversed else reaction.upper_bound
# subsystem
subsystem = check_none(reaction.subsystem.strip())
# get the model reaction
model_reaction_db = (session
.query(ModelReaction)
.filter(ModelReaction.reaction_id == reaction_db.id)
.filter(ModelReaction.model_id == model_db_id)
.filter(ModelReaction.lower_bound == lower_bound)
.filter(ModelReaction.upper_bound == upper_bound)
.filter(ModelReaction.gene_reaction_rule == reaction.gene_reaction_rule)
.filter(ModelReaction.objective_coefficient == reaction.objective_coefficient)
.filter(ModelReaction.subsystem == subsystem)
.first())
if model_reaction_db is None:
# get the number of existing copies of this reaction in the model
copy_number = (session
.query(ModelReaction)
.filter(ModelReaction.reaction_id == reaction_db.id)
.filter(ModelReaction.model_id == model_db_id)
.count()) + 1
# make a new reaction
model_reaction_db = ModelReaction(model_id=model_db_id,
reaction_id=reaction_db.id,
gene_reaction_rule=reaction.gene_reaction_rule,
original_gene_reaction_rule=reaction.gene_reaction_rule,
upper_bound=upper_bound,
lower_bound=lower_bound,
objective_coefficient=reaction.objective_coefficient,
copy_number=copy_number,
subsystem=subsystem)
session.add(model_reaction_db)
session.commit()
# remember the changed ids
model_db_rxn_ids[reaction.id] = model_reaction_db.id
# add synonyms
#
# get the id from the published model
for old_bigg_id in old_reaction_ids[reaction.id]:
# add a synonym
synonym_db = (session
.query(Synonym)
.filter(Synonym.type == 'reaction')
.filter(Synonym.ome_id == reaction_db.id)
.filter(Synonym.synonym == old_bigg_id)
.filter(Synonym.data_source_id == data_source_id)
.first())
if synonym_db is None:
synonym_db = Synonym(type='reaction',
ome_id=reaction_db.id,
synonym=old_bigg_id,
data_source_id=data_source_id)
session.add(synonym_db)
session.commit()
# add OldIDSynonym
old_id_db = (session
.query(OldIDSynonym)
.filter(OldIDSynonym.type == 'model_reaction')
.filter(OldIDSynonym.ome_id == model_reaction_db.id)
.filter(OldIDSynonym.synonym_id == synonym_db.id)
.first())
if old_id_db is None:
old_id_db = OldIDSynonym(type='model_reaction',
ome_id=model_reaction_db.id,
synonym_id=synonym_db.id)
session.add(old_id_db)
session.commit()
return model_db_rxn_ids
def load_metabolites(session, model_id, model, compartment_names,
old_metabolite_ids):
"""Load the metabolites as components and model components.
Arguments:
---------
session: An SQLAlchemy session.
model_id: The database ID for the model.
model: The COBRApy model.
old_metabolite_ids: A dictionary where keys are new IDs and values are old
IDs for compartmentalized metabolites.
Returns
-------
comp_comp_db_ids: A dictionary where keys are the original compartmentalized
metabolite ids and the values are the database IDs for the compartmentalized
components.
final_metabolite_ids: A new dictionary where keys are original
compartmentalized metabolite IDs from the model and values are the new
compartmentalized metabolite IDs.
"""
comp_comp_db_ids = {}
final_metabolite_ids = {}
# only grab this once
data_source_id = get_or_create_data_source(session, 'old_bigg_id')
# get metabolite id duplicates
met_dups = load_tsv(settings.metabolite_duplicates)
def _check_metabolite_duplicates(bigg_id):
"""Return a new ID if there is a preferred ID, otherwise None."""
for row in met_dups:
if bigg_id in row[1:]:
return row[0]
return None
# for each metabolite in the model
for metabolite in model.metabolites:
metabolite_id = parse.remove_duplicate_tag(metabolite.id)
try:
component_bigg_id, compartment_bigg_id = parse.split_compartment(metabolite_id)
except Exception:
logging.error(('Could not find compartment for metabolite %s in'
'model %s' % (metabolite_id, model.id)))
continue
preferred = _check_metabolite_duplicates(component_bigg_id)
new_bigg_id = preferred if preferred else component_bigg_id
# look for the formula in these places
formula_fns = [lambda m: getattr(m, 'formula', None), # support cobra v0.3 and 0.4
lambda m: m.notes.get('FORMULA', None),
lambda m: m.notes.get('FORMULA1', None)]
# Cast to string, but not for None
strip_str_or_none = lambda v: str(v).strip() if v is not None else None
# Ignore the empty string
ignore_empty_str = lambda s: s if s != '' else None
# Use a generator for lazy evaluation
values = (ignore_empty_str(strip_str_or_none(formula_fn(metabolite)))
for formula_fn in formula_fns)
# Get the first non-null result. Otherwise _formula = None.
_formula = format_formula(next(filter(None, values), None))
# Check for non-valid formulas
if parse.invalid_formula(_formula):
logging.warning('Invalid formula %s for metabolite %s in model %s' % (_formula, metabolite_id, model.id))
_formula = None
# get charge
try:
charge = int(metabolite.charge)
# check for float charge
if charge != metabolite.charge:
logging.warning('Could not load charge {} for {} in model {}'
.format(metabolite.charge, metabolite_id, model.id))
charge = None
except Exception:
if hasattr(metabolite, 'charge') and metabolite.charge is not None:
logging.debug('Could not convert charge to integer for metabolite {} in model {}: {}'
.format(metabolite_id, model.id, metabolite.charge))
charge = None
# If there is no metabolite, add a new one.
metabolite_db = (session
.query(Component)
.filter(Component.bigg_id == new_bigg_id)
.first())
# if necessary, add the new metabolite, and keep track of the ID
new_name = scrub_name(getattr(metabolite, 'name', None))
if metabolite_db is None:
# make the new metabolite
metabolite_db = Component(bigg_id=new_bigg_id, name=new_name)
session.add(metabolite_db)
session.commit()
else:
# If the metabolite is not new, consider improving the descriptive name
improve_name(session, metabolite_db, new_name)
# add the deprecated id if necessary
if metabolite_db.bigg_id != component_bigg_id:
get_or_create(session, DeprecatedID, deprecated_id=component_bigg_id,
type='component', ome_id=metabolite_db.id)
# if there is no compartment, add a new one
compartment_db = (session
.query(Compartment)
.filter(Compartment.bigg_id == compartment_bigg_id)
.first())
if compartment_db is None:
try:
name = compartment_names[compartment_bigg_id]
except KeyError:
logging.warning('No name found for compartment %s' % compartment_bigg_id)
name = ''
compartment_db = Compartment(bigg_id=compartment_bigg_id, name=name)
session.add(compartment_db)
session.commit()
# if there is no compartmentalized compartment, add a new one
comp_component_db = (session
.query(CompartmentalizedComponent)
.filter(CompartmentalizedComponent.component_id == metabolite_db.id)
.filter(CompartmentalizedComponent.compartment_id == compartment_db.id)
.first())
if comp_component_db is None:
comp_component_db = CompartmentalizedComponent(component_id=metabolite_db.id,
compartment_id=compartment_db.id)
session.add(comp_component_db)
session.commit()
# remember for adding the reaction
comp_comp_db_ids[metabolite.id] = comp_component_db.id
final_metabolite_ids[metabolite.id] = '%s_%s' % (new_bigg_id, compartment_bigg_id)
# if there is no model compartmentalized compartment, add a new one
model_comp_comp_db = (session
.query(ModelCompartmentalizedComponent)
.filter(ModelCompartmentalizedComponent.compartmentalized_component_id == comp_component_db.id)
.filter(ModelCompartmentalizedComponent.model_id == model_id)
.first())
if model_comp_comp_db is None:
model_comp_comp_db = ModelCompartmentalizedComponent(model_id=model_id,
compartmentalized_component_id=comp_component_db.id,
formula=_formula,
charge=charge)
session.add(model_comp_comp_db)
session.commit()
else:
if model_comp_comp_db.formula is None:
model_comp_comp_db.formula = _formula
if model_comp_comp_db.charge is None:
model_comp_comp_db.charge = charge
session.commit()
# add synonyms
for old_bigg_id_c in old_metabolite_ids[metabolite.id]:
# Add Synonym and OldIDSynonym
synonym_db = (session
.query(Synonym)
.filter(Synonym.type == 'compartmentalized_component')
.filter(Synonym.ome_id == comp_component_db.id)
.filter(Synonym.synonym == old_bigg_id_c)
.filter(Synonym.data_source_id == data_source_id)
.first())
if synonym_db is None:
synonym_db = Synonym(type='compartmentalized_component',
ome_id=comp_component_db.id,
synonym=old_bigg_id_c,
data_source_id=data_source_id)
session.add(synonym_db)
session.commit()
old_id_db = (session
.query(OldIDSynonym)
.filter(OldIDSynonym.type == 'model_compartmentalized_component')
.filter(OldIDSynonym.ome_id == model_comp_comp_db.id)
.filter(OldIDSynonym.synonym_id == synonym_db.id)
.first())
if old_id_db is None:
old_id_db = OldIDSynonym(type='model_compartmentalized_component',
ome_id=model_comp_comp_db.id,
synonym_id=synonym_db.id)
session.add(old_id_db)
session.commit()
# Also add Synonym and OldIDSynonym for the universal metabolite
try:
new_style_id = parse.id_for_new_id_style(
parse.fix_legacy_id(old_bigg_id_c, use_hyphens=False),
is_metabolite=True
)
old_bigg_id_c_without_compartment = parse.split_compartment(new_style_id)[0]
except Exception as e:
logging.warning(e.message)
else:
synonym_db_2 = (session
.query(Synonym)
.filter(Synonym.type == 'component')
.filter(Synonym.ome_id == metabolite_db.id)
.filter(Synonym.synonym == old_bigg_id_c_without_compartment)
.filter(Synonym.data_source_id == data_source_id)
.first())
if synonym_db_2 is None:
synonym_db_2 = Synonym(type='component',
ome_id=metabolite_db.id,
synonym=old_bigg_id_c_without_compartment,
data_source_id=data_source_id)
session.add(synonym_db_2)
session.commit()
old_id_db = (session
.query(OldIDSynonym)
.filter(OldIDSynonym.type == 'model_compartmentalized_component')
.filter(OldIDSynonym.ome_id == model_comp_comp_db.id)
.filter(OldIDSynonym.synonym_id == synonym_db_2.id)
.first())
if old_id_db is None:
old_id_db = OldIDSynonym(type='model_compartmentalized_component',
ome_id=model_comp_comp_db.id,
synonym_id=synonym_db_2.id)
session.add(old_id_db)
session.commit()
return comp_comp_db_ids, final_metabolite_ids
def load_reactions(session, model_db_id, model, old_reaction_ids,
comp_comp_db_ids, final_metabolite_ids):
"""Load the reactions and stoichiometries into the model.
TODO if the reaction is already loaded, we need to check the stoichometry
has. If that doesn't match, then add a new reaction with an incremented ID
(e.g. ACALD_1)
Arguments
---------
session: An SQLAlchemy session.
model_db_id: The database ID for the model.
model: The COBRApy model.
old_reaction_ids: A dictionary where keys are new IDs and values are old IDs
for reactions.
comp_comp_db_ids: A dictionary where keys are the original compartmentalized
metabolite ids and the values are the database IDs for the compartmentalized
components.
final_metabolite_ids: A new dictionary where keys are original
compartmentalized metabolite IDs from the model and values are the new
compartmentalized metabolite IDs.
Returns
-------
A dictionary with keys for reaction BiGG IDs in the model and values for the
associated ModelReaction.id in the database.
"""
# only grab this once
data_source_id = get_or_create_data_source(session, 'old_bigg_id')
# get reaction hash_prefs
hash_prefs = load_tsv(settings.reaction_hash_prefs)
def _check_hash_prefs(a_hash, is_pseudoreaction):
"""Return the preferred BiGG ID for a_hash, or None."""
for row in hash_prefs:
marked_pseudo = len(row) > 2 and row[2] == 'pseudoreaction'
if row[0] == a_hash and marked_pseudo == is_pseudoreaction:
return row[1]
return None
# Generate reaction hashes, and find reactions in the same model in opposite
# directions.
reaction_hashes = {r.id: parse.hash_reaction(r, final_metabolite_ids)
for r in model.reactions}
reverse_reaction_hashes = {r.id: parse.hash_reaction(r, final_metabolite_ids, reverse=True)
for r in model.reactions}
reverse_reaction_hashes_rev = {v: k for k, v in six.iteritems(reverse_reaction_hashes)}
reactions_not_to_reverse = set()
for r_id, h in six.iteritems(reaction_hashes):
if h in reverse_reaction_hashes_rev:
reactions_not_to_reverse.add(r_id)
reactions_not_to_reverse.add(reverse_reaction_hashes_rev[h])
model_db_rxn_ids = {}
for reaction in model.reactions:
# Drop duplicates label
reaction_id = parse.remove_duplicate_tag(reaction.id)
# Get the reaction
reaction_db = (session
.query(Reaction)
.filter(Reaction.bigg_id == reaction_id)
.first())
# check for pseudoreaction
is_pseudoreaction = check_pseudoreaction(reaction_id)
# calculate the hash
reaction_hash = reaction_hashes[reaction.id]
hash_db = (session
.query(Reaction)
.filter(Reaction.reaction_hash == reaction_hash)
.filter(Reaction.pseudoreaction == is_pseudoreaction)
.first())
# If there wasn't a match for the forward hash, also check the reverse
# hash. Do not check reverse hash for reactions with both directions
# defined in the same model (e.g. SUCDi and FRD7).
if not hash_db and reaction.id not in reactions_not_to_reverse:
reverse_hash_db = (session
.query(Reaction)
.filter(Reaction.reaction_hash == reverse_reaction_hashes[reaction.id])
.filter(Reaction.pseudoreaction == is_pseudoreaction)
.first())
else:
reverse_hash_db = None
# bigg_id match hash match b==h pseudoreaction example function
# n n n first GAPD _new_reaction (1)
# n n y first EX_glc_e _new_reaction (1)
# y n n incorrect GAPD _new_reaction & increment (2)
# y n y incorrect EX_glc_e _new_reaction & increment (2)
# n y n GAPDH after GAPD reaction = hash_reaction (3a)
# n y y EX_glc__e after EX_glc_e reaction = hash_reaction (3a)
# y y n n ? reaction = hash_reaction (3a)
# y y n y ? reaction = hash_reaction (3a)
# y y y n second GAPD reaction = bigg_reaction (3b)
# y y y y second EX_glc_e reaction = bigg_reaction (3b)
# NOTE: only check pseudoreaction hash against other pseudoreactions
# 4a and 4b are 3a and 3b with a reversed reaction
def _find_new_incremented_id(session, original_id):
"""Look for a reaction bigg_id that is not already taken."""
new_id = increment_id(original_id)
while True:
# Check for existing and deprecated reaction ids
if (session.query(Reaction).filter(Reaction.bigg_id == new_id).first() is None and
not _is_deprecated_reaction_id(session, new_id)):
return new_id
new_id = increment_id(new_id)
# Check for a preferred ID in the preferences, based on the forward
# hash. Don't check the reverse hash in preferences.
preferred_id = _check_hash_prefs(reaction_hash, is_pseudoreaction)
# no reversed by default
is_reversed = False
is_new = False
# (0) If there is a preferred ID, make that the new ID, and increment any old IDs
if preferred_id is not None:
# if the reaction already matches, just continue
if hash_db is not None and hash_db.bigg_id == preferred_id:
reaction_db = hash_db
# otherwise, make the new reaction
else:
# if existing reactions match the preferred reaction find a new,
# incremented id for the existing match
preferred_id_db = session.query(Reaction).filter(Reaction.bigg_id == preferred_id).first()
if preferred_id_db is not None:
new_id = _find_new_incremented_id(session, preferred_id)
logging.warn('Incrementing database reaction {} to {} and prefering {} (from model {}) based on hash preferences'
.format(preferred_id, new_id, preferred_id, model.id))
preferred_id_db.bigg_id = new_id
session.commit()
# make a new reaction for the preferred_id
reaction_db = _new_reaction(session, reaction, preferred_id,
reaction_hash, model_db_id, model,
is_pseudoreaction, comp_comp_db_ids)
is_new = True
# (1) no bigg_id matches, no stoichiometry match or pseudoreaction, then
# make a new reaction
elif reaction_db is None and hash_db is None and reverse_hash_db is None:
# check that the id is not deprecated
if _is_deprecated_reaction_id(session, reaction.id):
logging.error(('Keeping bigg_id {} (hash {} - from model {}) '
'even though it is on the deprecated ID list. '
'You should add it to reaction-hash-prefs.txt')
.format(reaction_id, reaction_hash, model.id))
reaction_db = _new_reaction(session, reaction, reaction_id,
reaction_hash, model_db_id, model,
is_pseudoreaction, comp_comp_db_ids)
is_new = True
# (2) bigg_id matches, but not the hash, then increment the BIGG_ID
elif reaction_db is not None and hash_db is None and reverse_hash_db is None:
# loop until we find a non-matching find non-matching ID
new_id = _find_new_incremented_id(session, reaction.id)
logging.warn('Incrementing bigg_id {} to {} (from model {}) based on conflicting reaction hash'
.format(reaction_id, new_id, model.id))
reaction_db = _new_reaction(session, reaction, new_id,
reaction_hash, model_db_id, model,
is_pseudoreaction, comp_comp_db_ids)
is_new = True
# (3) but found a stoichiometry match, then use the hash reaction match.
elif hash_db is not None:
# WARNING TODO this requires that loaded metabolites always match on
# bigg_id, which should be the case.
# (3a)
if reaction_db is None or reaction_db.id != hash_db.id:
reaction_db = hash_db
# (3b) BIGG ID matches a reaction with the same hash, then just continue
else:
pass
# (4) but found a stoichiometry match, then use the hash reaction match.
elif reverse_hash_db is not None:
# WARNING TODO this requires that loaded metabolites always match on
# bigg_id, which should be the case.
# Remember to switch upper and lower bounds
is_reversed = True
logging.info('Matched {} to {} based on reverse hash'
.format(reaction_id, reverse_hash_db.bigg_id))
# (4a)
if reaction_db is None or reaction_db.id != reverse_hash_db.id:
reaction_db = reverse_hash_db
# (4b) BIGG ID matches a reaction with the same hash, then just continue
else:
pass
else:
raise Exception('Should not get here')
# If the reaction is not new, consider improving the descriptive name
if not is_new:
new_name = scrub_name(check_none(getattr(reaction, 'name', None)))
improve_name(session, reaction_db, new_name)
# Add reaction to deprecated ID list if necessary
if reaction_db.bigg_id != reaction_id:
get_or_create(session, DeprecatedID, deprecated_id=reaction_id,
type='reaction', ome_id=reaction_db.id)
# If the reaction is reversed, then switch upper and lower bound
lower_bound = -reaction.upper_bound if is_reversed else reaction.lower_bound
upper_bound = -reaction.lower_bound if is_reversed else reaction.upper_bound
# subsystem
subsystem = check_none(reaction.subsystem.strip())
# get the model reaction
model_reaction_db = (session
.query(ModelReaction)
.filter(ModelReaction.reaction_id == reaction_db.id)
.filter(ModelReaction.model_id == model_db_id)
.filter(ModelReaction.lower_bound == lower_bound)
.filter(ModelReaction.upper_bound == upper_bound)
.filter(ModelReaction.gene_reaction_rule == reaction.gene_reaction_rule)
.filter(ModelReaction.objective_coefficient == reaction.objective_coefficient)
.filter(ModelReaction.subsystem == subsystem)
.first())
if model_reaction_db is None:
# get the number of existing copies of this reaction in the model
copy_number = (session
.query(ModelReaction)
.filter(ModelReaction.reaction_id == reaction_db.id)
.filter(ModelReaction.model_id == model_db_id)
.count()) + 1
# make a new reaction
model_reaction_db = ModelReaction(model_id=model_db_id,
reaction_id=reaction_db.id,
gene_reaction_rule=reaction.gene_reaction_rule,
original_gene_reaction_rule=reaction.gene_reaction_rule,
upper_bound=upper_bound,
lower_bound=lower_bound,
objective_coefficient=reaction.objective_coefficient,
copy_number=copy_number,
subsystem=subsystem)
session.add(model_reaction_db)
session.commit()
# remember the changed ids
model_db_rxn_ids[reaction.id] = model_reaction_db.id
# add synonyms
#
# get the id from the published model
for old_bigg_id in old_reaction_ids[reaction.id]:
# add a synonym
synonym_db = (session
.query(Synonym)
.filter(Synonym.type == 'reaction')
.filter(Synonym.ome_id == reaction_db.id)
.filter(Synonym.synonym == old_bigg_id)
.filter(Synonym.data_source_id == data_source_id)
.first())
if synonym_db is None:
synonym_db = Synonym(type='reaction',
ome_id=reaction_db.id,
synonym=old_bigg_id,
data_source_id=data_source_id)
session.add(synonym_db)
session.commit()
# add OldIDSynonym
old_id_db = (session
.query(OldIDSynonym)
.filter(OldIDSynonym.type == 'model_reaction')
.filter(OldIDSynonym.ome_id == model_reaction_db.id)
.filter(OldIDSynonym.synonym_id == synonym_db.id)
.first())
if old_id_db is None:
old_id_db = OldIDSynonym(type='model_reaction',
ome_id=model_reaction_db.id,
synonym_id=synonym_db.id)
session.add(old_id_db)
session.commit()
return model_db_rxn_ids
def load_metabolites(session, model_id, model, compartment_names,
old_metabolite_ids):
"""Load the metabolites as components and model components.
Arguments:
---------
session: An SQLAlchemy session.
model_id: The database ID for the model.
model: The COBRApy model.
old_metabolite_ids: A dictionary where keys are new IDs and values are old
IDs for compartmentalized metabolites.
Returns
-------
comp_comp_db_ids: A dictionary where keys are the original compartmentalized
metabolite ids and the values are the database IDs for the compartmentalized
components.
final_metabolite_ids: A new dictionary where keys are original
compartmentalized metabolite IDs from the model and values are the new
compartmentalized metabolite IDs.
"""
comp_comp_db_ids = {}
final_metabolite_ids = {}
# only grab this once
data_source_id = get_or_create_data_source(session, 'old_bigg_id')
# get metabolite id duplicates
met_dups = load_tsv(settings.metabolite_duplicates)
def _check_metabolite_duplicates(bigg_id):
"""Return a new ID if there is a preferred ID, otherwise None."""
for row in met_dups:
if bigg_id in row[1:]:
return row[0]
return None
# for each metabolite in the model
for metabolite in model.metabolites:
metabolite_id = parse.remove_duplicate_tag(metabolite.id)
try:
component_bigg_id, compartment_bigg_id = parse.split_compartment(metabolite_id)
except Exception:
logging.error(('Could not find compartment for metabolite %s in'
'model %s' % (metabolite_id, model.id)))
continue
preferred = _check_metabolite_duplicates(component_bigg_id)
new_bigg_id = preferred if preferred else component_bigg_id
# look for the formula in these places
formula_fns = [lambda m: getattr(m, 'formula', None), # support cobra v0.3 and 0.4
lambda m: m.notes.get('FORMULA', None),
lambda m: m.notes.get('FORMULA1', None)]
# Cast to string, but not for None
strip_str_or_none = lambda v: str(v).strip() if v is not None else None
# Ignore the empty string
ignore_empty_str = lambda s: s if s != '' else None
# Use a generator for lazy evaluation
values = (ignore_empty_str(strip_str_or_none(formula_fn(metabolite)))
for formula_fn in formula_fns)
# Get the first non-null result. Otherwise _formula = None.
_formula = format_formula(next(filter(None, values), None))
# Check for non-valid formulas
if parse.invalid_formula(_formula):
logging.warn('Invalid formula %s for metabolite %s in model %s' % (_formula, metabolite_id, model.id))
_formula = None
# get charge
try:
charge = int(metabolite.charge)
# check for float charge
if charge != metabolite.charge:
logging.warn('Could not load charge {} for {} in model {}'
.format(metabolite.charge, metabolite_id, model.id))
charge = None
except Exception:
if hasattr(metabolite, 'charge') and metabolite.charge is not None:
logging.debug('Could not convert charge to integer for metabolite {} in model {}: {}'
.format(metabolite_id, model.id, metabolite.charge))
charge = None
# If there is no metabolite, add a new one.
metabolite_db = (session
.query(Component)
.filter(Component.bigg_id == new_bigg_id)
.first())
# if necessary, add the new metabolite, and keep track of the ID
new_name = scrub_name(getattr(metabolite, 'name', None))
if metabolite_db is None:
# make the new metabolite
metabolite_db = Component(bigg_id=new_bigg_id, name=new_name)
session.add(metabolite_db)
session.commit()
else:
# If the metabolite is not new, consider improving the descriptive name
improve_name(session, metabolite_db, new_name)
# add the deprecated id if necessary
if metabolite_db.bigg_id != component_bigg_id:
get_or_create(session, DeprecatedID, deprecated_id=component_bigg_id,
type='component', ome_id=metabolite_db.id)
# if there is no compartment, add a new one
compartment_db = (session
.query(Compartment)
.filter(Compartment.bigg_id == compartment_bigg_id)
.first())
if compartment_db is None:
try:
name = compartment_names[compartment_bigg_id]
except KeyError:
logging.warn('No name found for compartment %s' % compartment_bigg_id)
name = ''
compartment_db = Compartment(bigg_id=compartment_bigg_id, name=name)
session.add(compartment_db)
session.commit()
# if there is no compartmentalized compartment, add a new one
comp_component_db = (session
.query(CompartmentalizedComponent)
.filter(CompartmentalizedComponent.component_id == metabolite_db.id)
.filter(CompartmentalizedComponent.compartment_id == compartment_db.id)
.first())
if comp_component_db is None:
comp_component_db = CompartmentalizedComponent(component_id=metabolite_db.id,
compartment_id=compartment_db.id)
session.add(comp_component_db)
session.commit()
# remember for adding the reaction
comp_comp_db_ids[metabolite.id] = comp_component_db.id
final_metabolite_ids[metabolite.id] = '%s_%s' % (new_bigg_id, compartment_bigg_id)
# if there is no model compartmentalized compartment, add a new one
model_comp_comp_db = (session
.query(ModelCompartmentalizedComponent)
.filter(ModelCompartmentalizedComponent.compartmentalized_component_id == comp_component_db.id)
.filter(ModelCompartmentalizedComponent.model_id == model_id)
.first())
if model_comp_comp_db is None:
model_comp_comp_db = ModelCompartmentalizedComponent(model_id=model_id,
compartmentalized_component_id=comp_component_db.id,
formula=_formula,
charge=charge)
session.add(model_comp_comp_db)
session.commit()
else:
if model_comp_comp_db.formula is None:
model_comp_comp_db.formula = _formula
if model_comp_comp_db.charge is None:
model_comp_comp_db.charge = charge
session.commit()
# add synonyms
for old_bigg_id_c in old_metabolite_ids[metabolite.id]:
# Add Synonym and OldIDSynonym
synonym_db = (session
.query(Synonym)
.filter(Synonym.type == 'compartmentalized_component')
.filter(Synonym.ome_id == comp_component_db.id)
.filter(Synonym.synonym == old_bigg_id_c)
.filter(Synonym.data_source_id == data_source_id)
.first())
if synonym_db is None:
synonym_db = Synonym(type='compartmentalized_component',
ome_id=comp_component_db.id,
synonym=old_bigg_id_c,
data_source_id=data_source_id)
session.add(synonym_db)
session.commit()
old_id_db = (session
.query(OldIDSynonym)
.filter(OldIDSynonym.type == 'model_compartmentalized_component')
.filter(OldIDSynonym.ome_id == model_comp_comp_db.id)
.filter(OldIDSynonym.synonym_id == synonym_db.id)
.first())
if old_id_db is None:
old_id_db = OldIDSynonym(type='model_compartmentalized_component',
ome_id=model_comp_comp_db.id,
synonym_id=synonym_db.id)
session.add(old_id_db)
session.commit()
# Also add Synonym and OldIDSynonym for the universal metabolite
try:
new_style_id = parse.id_for_new_id_style(
parse.fix_legacy_id(old_bigg_id_c, use_hyphens=False),
is_metabolite=True
)
old_bigg_id_c_without_compartment = parse.split_compartment(new_style_id)[0]
except Exception as e:
logging.warn(e.message)
else:
synonym_db_2 = (session
.query(Synonym)
.filter(Synonym.type == 'component')
.filter(Synonym.ome_id == metabolite_db.id)
.filter(Synonym.synonym == old_bigg_id_c_without_compartment)
.filter(Synonym.data_source_id == data_source_id)
.first())
if synonym_db_2 is None:
synonym_db_2 = Synonym(type='component',
ome_id=metabolite_db.id,
synonym=old_bigg_id_c_without_compartment,
data_source_id=data_source_id)
session.add(synonym_db_2)
session.commit()
old_id_db = (session
.query(OldIDSynonym)
.filter(OldIDSynonym.type == 'model_compartmentalized_component')
.filter(OldIDSynonym.ome_id == model_comp_comp_db.id)
.filter(OldIDSynonym.synonym_id == synonym_db_2.id)
.first())
if old_id_db is None:
old_id_db = OldIDSynonym(type='model_compartmentalized_component',
ome_id=model_comp_comp_db.id,
synonym_id=synonym_db_2.id)
session.add(old_id_db)
session.commit()
return comp_comp_db_ids, final_metabolite_ids