def __init__(self,
community,
slim=False,
reactions=None,
metabolites=None):
"""Get the solution from a community model."""
if reactions is None:
reactions = community.reactions
if metabolites is None:
metabolites = community.metabolites
rids = np.array([(r.global_id, r.community_id) for r in reactions])
mids = np.array([(m.global_id, m.community_id) for m in metabolites])
if not slim:
var_primals = community.solver.primal_values
fluxes = pd.Series(
[
var_primals[r.id] - var_primals[r.reverse_id]
for r in reactions
],
name="fluxes",
)
super(CommunitySolution, self).__init__(
community.solver.objective.value,
community.solver.status,
_group_taxa(fluxes, rids[:, 0], rids[:, 1]),
None,
None,
)
else:
super(CommunitySolution, self).__init__(
community.solver.objective.value,
community.solver.status,
None,
None,
None,
)
gcs = pd.Series()
for sp in community.taxa:
gcs[sp] = community.constraints["objective_" + sp].primal
# Workaround for an optlang bug (PR #120)
if interface_to_str(community.problem) == "gurobi":
gcs = gcs.abs()
self.strategy = community.modification
self.members = pd.DataFrame({
"abundance":
community.abundances,
"growth_rate":
gcs,
"reactions":
pd.Series(Counter(rids[:, 1])),
"metabolites":
pd.Series(Counter(mids[:, 1])),
})
self.members.index.name = "compartments"
self.growth_rate = sum(community.abundances * gcs)
# Save estimated accuracy for osqp
if interface_to_str(community.problem) == "osqp":
self.primal_residual = community.solver.problem.info.pri_res
self.dual_residual = community.solver.problem.info.dua_res
def cooperative_tradeoff(community, min_growth, fraction, fluxes, pfba):
"""Find the best tradeoff between community and individual growth."""
with community as com:
solver = interface_to_str(community.problem)
check_modification(community)
min_growth = _format_min_growth(min_growth, community.taxa)
_apply_min_growth(community, min_growth)
com.objective = com.scale * com.variables.community_objective
min_growth = (optimize_with_retry(
com, message="could not get community growth rate.") / com.scale)
if not isinstance(fraction, Sized):
fraction = [fraction]
else:
fraction = np.sort(fraction)[::-1]
# Add needed variables etc.
regularize_l2_norm(com, 0.0)
results = []
for fr in fraction:
com.variables.community_objective.lb = fr * min_growth
com.variables.community_objective.ub = min_growth
sol = solve(community, fluxes=fluxes, pfba=pfba)
# OSQP is better with QPs then LPs
# so it won't get better with the crossover
if (sol.status != OPTIMAL and solver != "osqp"):
sol = crossover(com, sol, fluxes=fluxes, pfba=pfba)
results.append((fr, sol))
if len(results) == 1:
return results[0][1]
return pd.DataFrame.from_records(results,
columns=["tradeoff", "solution"])
def reset_solver(community):
"""Reset the solver."""
interface = interface_to_str(community.solver.interface)
logger.info("resetting solver, hoping for the best.")
if interface == "cplex":
logger.warning("switching cplex LP algorithm to `network`.")
community.solver.configuration.lp_method = "network"
elif interface == "gurobi":
community.solver.problem.reset()
elif interface == "glpk":
glp_adv_basis(community.solver.problem, 0)
def reset_solver(community):
"""Reset the solver."""
interface = interface_to_str(community.solver.interface)
logger.info("resetting solver, hoping for the best.")
if interface == "cplex":
community.solver.configuration.lp_method = "network"
community.solver.configuration.lp_method = "barrier"
elif interface == "gurobi":
community.solver.problem.reset()
elif interface == "glpk":
glp_adv_basis(community.solver.problem, 0)
elif interface == "osqp":
community.solver.problem.reset()