What is the version requirement for swiglpk?

[jonrkarr]

I see errors with cbmpy.CBGLPK.glpk_FluxVariabilityAnalysis in 0.8.1 and swiglpk 4.65.1, 5.0.0, and 5.0.3 from PyPI. Is a newer version of swiglpk needed?

>       solution = cbmpy.CBGLPK.glpk_FluxVariabilityAnalysis(model)

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
../../.local/lib/python3.9/site-packages/cbmpy/CBGLPK.py:677: in glpk_FluxVariabilityAnalysis
    glpk_func_SetObjectiveFunctionAsConstraint(
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

cpx = <Swig Object of type 'glp_prob *' at 0x7f3a5dc5d060>, rhs_sense = 'lower', oval = 0.8739215069684909, tol = None, optPercentage = 100.0

    def glpk_func_SetObjectiveFunctionAsConstraint(
        cpx, rhs_sense, oval, tol, optPercentage
    ):
        """
        Take the objective function and "optimum" value and add it as a constraint
         - *cpx* a cplex object
         - *oval* the objective value
         - *tol*  [default=None] do not floor/ceiling the objective function constraint, otherwise round of to *tol*
         - *rhs_sense* [default='lower'] means objC >= objVal the inequality to use for the objective constraint can also be *upper* or *equal*
         - *optPercentage* [default=100.0] means the percentage optimal value to use for the RHS of the objective constraint: optimal_value*(optPercentage/100.0)
    
        """
    
        # generate new constraint from old objective value (use non-zero coefficients)
        LCS = []
        LCN = []
        new_constraint = []
>       for v_ in range(len(cpx.cols)):
E       AttributeError: 'SwigPyObject' object has no attribute 'cols'

../../.local/lib/python3.9/site-packages/cbmpy/CBGLPK.py:943: AttributeError

[jonrkarr]

Below are steps to repeat the error:

Run a Docker image

docker run -it --rm --entrypoint bash python:3.9-slim-buster

In the Docker image

# get example model
apt-get update -y
apt-get install -y wget
wget https://raw.githubusercontent.com/biosimulators/Biosimulators_CBMPy/dev/tests/fixtures/textbook.xml

# install cbmpy
pip install cbmpy swiglpk sympy python_libsbml lxml

# launch python
python

In the python shell:

import cbmpy
import cbmpy.CBGLPK
model_filename = 'textbook.xml'
model = cbmpy.CBRead.readSBML3FBC(model_filename)
solution = cbmpy.CBGLPK.glpk_FluxVariabilityAnalysis(model)

This terminates with this error:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.9/site-packages/cbmpy/CBGLPK.py", line 677, in glpk_FluxVariabilityAnalysis
    glpk_func_SetObjectiveFunctionAsConstraint(
  File "/usr/local/lib/python3.9/site-packages/cbmpy/CBGLPK.py", line 943, in glpk_func_SetObjectiveFunctionAsConstraint
    for v_ in range(len(cpx.cols)):
AttributeError: 'SwigPyObject' object has no attribute 'cols'

[bgoli]

The pyGLPK version that is currently used is not available for Python 3 so CPLEX is is required ... I am working on a new wrapper that will use the swiglpk used by cobrapy which should resolve all compatibility issues.