GrowthCouplingPotential

[KristianJensen]

aka OptCouple This PR adds my growth coupling potential based MILP optimization to the strain_design.deterministic.linear_programming module I’ve also added some tests, but I don’t really know how pytest works, so I’ve just tried to mirror the OptKnock tests

[codecov-io]

Codecov Report

Merging #215 (80cffeb) into devel (598b0a0) will decrease coverage by 0.75%. The diff coverage is 43.60%.

:exclamation: Current head 80cffeb differs from pull request most recent head 9451a04. Consider uploading reports for the commit 9451a04 to get more accurate results

@@            Coverage Diff             @@
##            devel     #215      +/-   ##
==========================================
- Coverage   64.11%   63.35%   -0.76%     
==========================================
  Files          58       58              
  Lines        5576     5786     +210     
  Branches      949      997      +48     
==========================================
+ Hits         3575     3666      +91     
- Misses       1795     1902     +107     
- Partials      206      218      +12     

Impacted Files	Coverage Δ
...ain_design/deterministic/flux_variability_based.py	63.69% <ø> (ø)
.../strain_design/deterministic/linear_programming.py	32.88% <43.60%> (+9.37%)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 598b0a0...9451a04. Read the comment docs.

[bdelepine]

Hello!

Do you have any update on this PR? Is it still alive? Is it safe to use this branch to try out OptCouple?

[KristianJensen]

Hi @bdelepine

The PR is a bit stale as I haven't had time to finalize the unittests and incorporate the suggested changes. But the method should work. If you encounter any problems, feel free to post them here.

EDIT: Looking at the comments again, it does actually look like there have been some changes to cobrapy in the meantime that will cause this branch not to work. I will try to fix this ASAP, as I would also like to get this merged.

[bdelepine]

Hi @KristianJensen,

Sadly, I confirm that the current branch does not work. The sol.f triggered an AttributeError. Once corrected, the code runs (using CPLEX). Unfortunately, I was still unable to get results since I keep getting {'knockouts': [], 'knockins': [], 'medium': [], 'obj_val': 0.0} despite trying different set of parameters. I guess my example is broken or that something is silently off somewhere 😕.

HTH, +

[KristianJensen]

Thanks for the report @bdelepine I will try to get this PR finished ASAP - I will also need to use it myself in my current work. I have now addressed some of @Midnighter's comments and updated the code to be compatible with modern cobrapy. I will do some testing of the method next.

If you want to try the method in the meantime, there is an implementation in the supplementary material of the paper, including a list of package versions.

[KristianJensen]

@Midnighter I have pushed some changes:

• The code is now compatible with modern cobrapy
• The test no longer relies on gurobi, but runs with glpk
• I've changed the test to a target that is not already growth-coupled (and added an assertion to verify this)

Let me know if there are other issues I should address (or if I should ping someone else instead :) )

[phantomas1234]

Thanks @carrascomj!