def find_gene_knockout_reactions(cobra_model,
gene_list,
compiled_gene_reaction_rules=None):
"""identify reactions which will be disabled when the genes are knocked out
cobra_model: :class:`~cobra.core.Model.Model`
gene_list: iterable of :class:`~cobra.core.Gene.Gene`
compiled_gene_reaction_rules: dict of {reaction_id: compiled_string}
If provided, this gives pre-compiled gene_reaction_rule strings.
The compiled rule strings can be evaluated much faster. If a rule
is not provided, the regular expression evaluation will be used.
Because not all gene_reaction_rule strings can be evaluated, this
dict must exclude any rules which can not be used with eval.
"""
potential_reactions = set()
for gene in gene_list:
if isinstance(gene, string_types):
gene = cobra_model.genes.get_by_id(gene)
potential_reactions.update(gene._reaction)
gene_set = {str(i) for i in gene_list}
if compiled_gene_reaction_rules is None:
compiled_gene_reaction_rules = {
r: parse_gpr(r.gene_reaction_rule)[0]
for r in potential_reactions
}
return [
r for r in potential_reactions
if not eval_gpr(compiled_gene_reaction_rules[r], gene_set)
]
def validate_model(model):
errors = []
warnings = []
errors.extend(check_reaction_bounds(model))
errors.extend(check_metabolite_compartment_formula(model))
# test gpr
for reaction in model.reactions:
try:
parse_gpr(reaction.gene_reaction_rule)
except SyntaxError:
errors.append("reaction '%s' has invalid gpr '%s'" %
(reaction.id, reaction.gene_reaction_rule))
# test mass balance
for reaction, balance in iteritems(check_mass_balance(model)):
# check if it's a demand or exchange reaction
if len(reaction.metabolites) == 1:
warnings.append("reaction '%s' is not balanced. Should it "
"be annotated as a demand or exchange "
"reaction?" % reaction.id)
elif "biomass" in reaction.id.lower():
warnings.append("reaction '%s' is not balanced. Should it "
"be annotated as a biomass reaction?" %
reaction.id)
else:
warnings.append("reaction '%s' is not balanced for %s" %
(reaction.id, ", ".join(sorted(balance))))
# try solving
solution = model.optimize(solver="esolver")
if solution.status != "optimal":
errors.append("model can not be solved (status '%s')" %
solution.status)
return {"errors": errors, "warnings": warnings}
# if there is no objective, then we know why the objective was low
if len(model.objective) == 0:
warnings.append("model has no objective function")
elif solution.f <= 0:
warnings.append("model can not produce nonzero biomass")
elif solution.f <= 1e-3:
warnings.append("biomass flux %s too low" % str(solution.f))
if len(model.objective) > 1:
warnings.append("model should only have one reaction as the objective")
return {"errors": errors, "warnings": warnings, "objective": solution.f}
def validate_model(model):
errors = []
warnings = []
errors.extend(check_reaction_bounds(model))
errors.extend(check_metabolite_compartment_formula(model))
# test gpr
for reaction in model.reactions:
try:
parse_gpr(reaction.gene_reaction_rule)
except SyntaxError:
errors.append("reaction '%s' has invalid gpr '%s'" %
(reaction.id, reaction.gene_reaction_rule))
# test mass balance
for reaction, balance in iteritems(check_mass_balance(model)):
# check if it's a demand or exchange reaction
if len(reaction.metabolites) == 1:
warnings.append("reaction '%s' is not balanced. Should it "
"be annotated as a demand or exchange "
"reaction?" % reaction.id)
elif "biomass" in reaction.id.lower():
warnings.append("reaction '%s' is not balanced. Should it "
"be annotated as a biomass reaction?" %
reaction.id)
else:
warnings.append("reaction '%s' is not balanced for %s" %
(reaction.id, ", ".join(sorted(balance))))
# try solving
solution = model.optimize(solver="esolver")
if solution.status != "optimal":
errors.append("model can not be solved (status '%s')" %
solution.status)
return {"errors": errors, "warnings": warnings}
# if there is no objective, then we know why the objective was low
if len(model.objective) == 0:
warnings.append("model has no objective function")
elif solution.f <= 0:
warnings.append("model can not produce nonzero biomass")
elif solution.f <= 1e-3:
warnings.append("biomass flux %s too low" % str(solution.f))
if len(model.objective) > 1:
warnings.append("model should only have one reaction as the objective")
return {"errors": errors, "warnings": warnings, "objective": solution.f}
def reaction_confidence(rule, conf_genes):
"""Calculates the confidence for the reaction based on a gene-reaction
rule.
Args:
rule (str): A gene-reaction rule. For instance "A and B".
conf_genes (dict): A str->int map denoting the mapping of gene IDs
to expression confidence values. Allowed confidence values are -1
(absent/do not include), 0 (unknown), 1 (low confidence),
2 (medium confidence) and 3 (high confidence).
"""
ast_rule, _ = parse_gpr(rule)
return safe_eval_gpr(ast_rule, conf_genes)
def get_compiled_gene_reaction_rules(cobra_model):
"""Generates a dict of compiled gene_reaction_rules
Any gene_reaction_rule expressions which cannot be compiled or do not
evaluate after compiling will be excluded. The result can be used in the
find_gene_knockout_reactions function to speed up evaluation of these
rules.
"""
return {
r: parse_gpr(r.gene_reaction_rule)[0]
for r in cobra_model.reactions
}
pheno_ess_reac_dict[reac].append(scr_id)
# print "Reaction {} is an essential reaction for {}.".format(scr_id, reac)
elif test_stat == 'optimal':
vKO = math.fabs(test_f)
vOrg = math.fabs(org_pheno_f)
if vKO <= vOrg * ess_thr:
pheno_ess_reac_dict[reac].append(scr_id)
# print "Reaction {} is an essential reaction for {}.".format(scr_id, reac)
"""Extract the related genes in the essential reactions as the essential genes of the phenotypic data"""
pheno_ess_gene_dict = defaultdict(list)
ess_gene_pheno_dict = defaultdict(list)
for keys, values in pheno_ess_reac_dict.items():
print "Reaction {} has {} essential reactions.".format(keys, len(values))
for ess_reac_id in values:
ess_reac = model.reactions.get_by_id(ess_reac_id)
temp_gpr = parse_gpr(ess_reac.gene_reaction_rule)
# print len(temp_gpr[1])
for z in temp_gpr[1]:
if z not in pheno_ess_gene_dict[keys] and z not in pseudo_gene_rec:
pheno_ess_gene_dict[keys].append(z)
ess_gene_pheno_dict[z].append(keys)
print "Reaction {} has {} essential genes.\n".format(
keys, len(pheno_ess_gene_dict[keys]))
# pheno_ess_bool_dict = defaultdict(list)
# ess_pheno_bool_dict = defaultdict(list)
"""Generate the final output protein-phenotype masking matrix."""
row_num = len(phenotype_id_rec)
col_num = len(gene_id_rec)
mask_matrix = np.zeros((row_num, col_num), dtype=np.int)
pheno_id = pheno_ess_gene_dict.keys()