[BUG] check_mass_balance fails for floating point coefficients

[jedforrest]

Is there an existing issue for this?

• I have searched the existing issues

Problem description

The check_mass_balance function for reactions gives an incorrect result for reactions with floating point coefficients.

Specifically, scaling the coefficients by a float (like in the example below) returns a dictionary of near-zero values. A non-empty dictionary implies the reaction is unbalanced, when it's really just a floating-point error.

A simple fix is to filter out near-zero values with something like numpy.isclose.

Code sample

Code run:

from cobra import Model, Metabolite, Reaction

# setup
model = Model("model")

adp = Metabolite("adp", formula="C10H12N5O10P2", charge=-3)
atp = Metabolite("atp", formula="C10H12N5O13P3", charge=-4)
h = Metabolite("h", formula="H", charge=1)
h2o = Metabolite("h2o", formula="H2O", charge=0)
pi = Metabolite("pi", formula="HO4P", charge=-2)
model.add_metabolites([adp, atp, h, h2o, pi])

reaction = Reaction("reaction")
reaction.add_metabolites({atp: -1, h2o: -1, adp: 1, pi: 1, h: 1})
model.add_reactions([reaction])

print(reaction.check_mass_balance())
# {}

# float coefficients return a mass balance dict of near-zero values
reaction_scaled = reaction * 1.3
print(reaction_scaled.check_mass_balance())
# {'charge': -2.220446049250313e-16, 'H': -1.3322676295501878e-15, 'O': -2.6645352591003757e-15, 'P': -2.220446049250313e-16}

Environment

Anything else?

No response