Incorrect size(model.rhs) error when running loopless FVA with LLC-NS method

[assaron]

I'm trying to use LLC-NS constraints for loopless FVA and it fails with an error, while fastSNP method works well.

Here is the code I'm running:

model = readCbModel('e_coli_core.mat');

res = fluxVariability(model, 100, 'max', model.rxns, 1, true);
res = fluxVariability(model, 100, 'max', model.rxns, 1, 'LLC-NS');
res = fluxVariability(model, 100, 'max', model.rxns, 1, 'fastSNP');

The first and third fluxVariability calls work without error, but for the second I get:

Reduce complexity by nullspace preprocessing and implementing localized loopless constraints (LLCs)
Reactions in internal nullspace can be divided into 1 connected components.
Error using gurobi
Incorrect size(model.rhs)

Error in solveCobraLP (line 839)
        resultgurobi = gurobi(gurobiLP,gurobiParam);
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Error in optimizeCbModel (line 652)
            solution = solveCobraLP(optProblem2, param);
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Error in processingLLCs>preprocessLLCs (line 151)
sFeas = optimizeCbModel(model2, 'max', 'one');
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Error in processingLLCs (line 36)
        loopInfo = preprocessLLCs(varargin{1:5});
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Error in fluxVariability (line 232)
        [solveLP, MILPproblem, loopInfo] = processingLLCs('preprocess', loopInfo, LPproblem, model, nRxns, osenseStr, MILPproblem);
                                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

I've added a breakpoint at solveCobraLP and for some reason the problem has matrix A of size 263x285 while vector b is of size 262, which causes the gurobi error.

Also, when I ran testFVA I also have the same error.

I hereby confirm that I have:

• Tried to solve the issue on my own
• Retried to run my code with the latest version of The COBRA Toolbox
• Checked that a similar issue has not already been opened

(Note: You may replace [ ] with [X] to check the box)

[rmtfleming]

Its probably added a constraint from the objective to make sure solutions are optimal. Evidently the rhs needs an extra row. c'*v = rhs(end+1) = c'*v_star where v_star is the optimal value of the LP objective. Pls make the edit and pr

…

On Sun, 10 Nov 2024 at 17:39, Alexey Sergushichev ***@***.***> wrote: I'm trying to use LLC-NS constraints for loopless FVA and it fails with an error, while fastSNP method works well. Here is the code I'm running: model = readCbModel('e_coli_core.mat'); res = fluxVariability(model, 100, 'max', model.rxns, 1, true); res = fluxVariability(model, 100, 'max', model.rxns, 1, 'LLC-NS'); res = fluxVariability(model, 100, 'max', model.rxns, 1, 'fastSNP'); The first and third fluxVariability calls work without error, but for the second I get: Reduce complexity by nullspace preprocessing and implementing localized loopless constraints (LLCs) Reactions in internal nullspace can be divided into 1 connected components. Error using gurobi Incorrect size(model.rhs) Error in solveCobraLP (line 839) resultgurobi = gurobi(gurobiLP,gurobiParam); ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Error in optimizeCbModel (line 652) solution = solveCobraLP(optProblem2, param); ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Error in processingLLCs>preprocessLLCs (line 151) sFeas = optimizeCbModel(model2, 'max', 'one'); ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Error in processingLLCs (line 36) loopInfo = preprocessLLCs(varargin{1:5}); ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Error in fluxVariability (line 232) [solveLP, MILPproblem, loopInfo] = processingLLCs('preprocess', loopInfo, LPproblem, model, nRxns, osenseStr, MILPproblem); ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ I've added a breakpoint at solveCobraLP and for some reason the problem has matrix A of size 263x285 while vector b is of size 262, which causes the gurobi error. Also, when I ran testFVA I also have the same error. *I hereby confirm that I have:* - Tried to solve the issue on my own - Retried to run my code with the latest version of The COBRA Toolbox - Checked that a similar issue has not already been opened *(Note: You may replace [ ] with [X] to check the box)* — Reply to this email directly, view it on GitHub <#2377>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAQMEOUYBZI56YDYVYIC6CLZ76K4NAVCNFSM6AAAAABRQNRJQCVHI2DSMVQWIX3LMV43ASLTON2WKOZSGY2DONBWHEYDMMY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- -- Mr. Ronan MT Fleming B.V.M.S. Dip. Math. Ph.D. ---------------------------------------------------------------------------- Associate Professor, Digital Metabolic Twin Center, www.digitalmetabolictwin.org School of Medicine, University of Galway. Ireland. ---------------------------------------------------------------------------- Coordinator of the Horizon Europe project "Reconstruction and Computational Modelling for Inherited Metabolic Diseases" (Recon4IMD <https://www.recon4imd.org/>) ---------------------------------------------------------------------------- Peer-reviewed publications: https://goo.gl/FZPG23 Mobile: +353 852 109 806 Github: @rmtfleming <https://github.com/rmtfleming> ----------------------------------------------------------------------------

[assaron]

Sorry, I'm not that familiar for cobra code base, neither with matlab in general, so not sure what you suggest exactly.

That said, I agree with your "evidently". As I understand at this line

cobratoolbox/src/analysis/FBA/optimizeCbModel.m

Line 636 in 1b4e503

optProblem2.b = [optProblem.b;zeros(2*nRxns,1);objectiveLP];

objectiveLP is empty and doesn't increase b size, while the constraint rows is added to A (although it's a zero row in my case).

I've also checked version 9d88036. There everything appears to work fine.

[assaron]

I've also found the commit that introduced this problem: 27d920b#diff-0401bfa37ea7edab5d85374c70147dfc317f57bcc6c1fbe031f112898b3ce8e2L438

I would suggest to revert to setting objectiveLP = 0;, but not sure why it was changed in the first place.

[rmtfleming]

OK, try switching https://github.com/opencobra/cobratoolbox/blob/develop/src/analysis/FBA/optimizeCbModel.m#L474C5-L474C16 to objectiveLP = 0;

…

On Mon, 11 Nov 2024 at 01:57, Alexey Sergushichev ***@***.***> wrote: I've also found the commit that introduced this problem: 27d920b #diff-0401bfa37ea7edab5d85374c70147dfc317f57bcc6c1fbe031f112898b3ce8e2L438 <27d920b#diff-0401bfa37ea7edab5d85374c70147dfc317f57bcc6c1fbe031f112898b3ce8e2L438> I would suggest to revert to setting objectiveLP = 0;, but not sure why it was changed in the first place. — Reply to this email directly, view it on GitHub <#2377 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAQMEOTDURXR2DLEENYEHFD2AAFH5AVCNFSM6AAAAABRQNRJQCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDINRXGA4DEMRSHE> . You are receiving this because you were mentioned.Message ID: ***@***.***>

-- -- Mr. Ronan MT Fleming B.V.M.S. Dip. Math. Ph.D. ---------------------------------------------------------------------------- Associate Professor, Digital Metabolic Twin Center, www.digitalmetabolictwin.org School of Medicine, University of Galway. Ireland. ---------------------------------------------------------------------------- Coordinator of the Horizon Europe project "Reconstruction and Computational Modelling for Inherited Metabolic Diseases" (Recon4IMD <https://www.recon4imd.org/>) ---------------------------------------------------------------------------- Peer-reviewed publications: https://goo.gl/FZPG23 Mobile: +353 852 109 806 Github: @rmtfleming <https://github.com/rmtfleming> ----------------------------------------------------------------------------

[assaron]

Actually, the problem seems to be fixed in the develop branch by this check: https://github.com/opencobra/cobratoolbox/blob/develop/src/analysis/FBA/optimizeCbModel.m#L636