Bug report: by checkInstallation, gurobi check fails from v2.10.5 onward; HMMER check fails from v2.11.0 onward.

[dimwas]

Hi,

I cloned the last version of RAVEN and noticed that the checks for gurobi and HMMER have failed during checkInstallation. I do not use HMMER, but I do use gurobi, and in both Python and COBRA it works good. After many attempts to fix gurobi, I decided to check the previous versions of RAVEN. It turned out, RAVEN versions before 2.10.5 do not throw failed checks in checkInstallation on either gurobi or HMMER. Starting from 2.11.0, both checks are failed.
Please take a look on it.

Best,
Wassili

[edkerk]

What MATLAB and operating system do you use? Can you post the whole checkInstallation output?
Have you tried running setRavenSolver('gurobi') and solveLP() anyway (see #602)?

[dimwas]

1. MATLAB 2025b.
2. Windows 11.

>> checkInstallation

*** THE RAVEN TOOLBOX ***

 > Installation type                    Advanced (via git)
 > Installing from location             C:\RAVEN
 > Checking RAVEN release               2.11.3
   > You are running the latest RAVEN release
 > Checking MATLAB release              2025b
 > Checking system architecture         win64
 > Set RAVEN in MATLAB path             Pass
 > Save MATLAB path                     Pass
 > Store RAVEN path as MATLAB pref      Pass

=== Model import and export ===
 > Add Java paths for Excel format      Pass
 > Checking libSBML version             5.20.2
 > Checking model import and export
   > Import Excel format                Pass
   > Export Excel format                Pass
   > Import SBML format                 Pass
   > Export SBML format                 Pass
   > Import YAML format                 Pass
   > Export YAML format                 Pass

=== Model solvers ===
 > Checking for LP solvers
   > glpk                               Pass
   > gurobi                             Fail
   > scip                               Fail
   > cobra                              Pass
 > Set RAVEN solver                     glpk

=== Essential binary executables ===
 > Checking BLAST+                      Pass
 > Checking DIAMOND                     Pass
 > Checking HMMER                       Fail
   Download/compile the binary and rerun checkInstallation
   This is essential to run getKEGGModelFromHomology()
   when using a FASTA file as input

=== Compatibility ===
 > Checking function uniqueness         Pass

*** checkInstallation complete ***

ans =

    '2.11.3'

4. Yes, it doesn't work with the latest RAVEN version:

>> solveLP(model)
getCurrentTask requires [Parallel Computing Toolbox](matlab:matlab.internal.addons.launchers.showExplorer('ErrorRecovery', 'identifier', 'DM', 'focused', 'getCurrentTask');).

Error in [optimizeProb](matlab:matlab.lang.internal.introspective.errorDocCallback('optimizeProb', 'C:\RAVEN\solver\optimizeProb.m', 93)) ([line 93](matlab: opentoline('C:\RAVEN\solver\optimizeProb.m',93,0)))
        if ~isempty(getCurrentTask) % If run in parallel, then one thread per gurobi
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Error in [solveLP](matlab:matlab.lang.internal.introspective.errorDocCallback('solveLP', 'C:\RAVEN\solver\solveLP.m', 104)) ([line 104](matlab: opentoline('C:\RAVEN\solver\solveLP.m',104,0)))
res = optimizeProb(prob,params);
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

, but works with v.2.10.4:

>> solveLP(model)

ans = 

  [struct](matlab:helpPopup('struct')) with fields:

         x: [1455×1 double]
         f: 1.3602
      stat: 1
       msg: 'Optimal solution found'
    sPrice: [1335×1 double]
     rCost: [1455×1 double]

[edkerk]

Thank you, it's a different bug than what we've seen before. Working on a fix right now.

[edkerk]

Please check if #753 solves it.

[dimwas]

The problem with Gurobi is solved, solveLP works. The problem with HMMER persists; now checkInstallation suggests to install WLS, restart and try again. I did as suggested, and nothing changed. I did not "Download/compile the binary and rerun checkInstallation", because I do not use HMMER and do not know what is meant by it.

>> checkInstallation

*** THE RAVEN TOOLBOX ***

 > Installation type                    Advanced (via git)
 > Installing from location             C:\RAVEN
 > Checking RAVEN release               DEVELOPMENT
 > Checking MATLAB release              2025b
 > Checking system architecture         win64
 > Set RAVEN in MATLAB path             Pass
 > Save MATLAB path                     Pass
 > Store RAVEN path as MATLAB pref      Pass

=== Model import and export ===
 > Add Java paths for Excel format      Pass
 > Checking libSBML version             5.20.2
 > Checking model import and export
   > Import Excel format                Pass
   > Export Excel format                Pass
   > Import SBML format                 Pass
   > Export SBML format                 Pass
   > Import YAML format                 Pass
   > Export YAML format                 Pass

=== Model solvers ===
 > Checking for LP solvers
   > glpk                               Pass
   > gurobi                             Pass
   > scip                               Fail
   > cobra                              Pass
 > Set RAVEN solver                     gurobi

=== Essential binary executables ===
 > Checking BLAST+                      Pass
 > Checking DIAMOND                     Pass
 > Checking HMMER                       Fail
   Download/compile the binary and rerun checkInstallation
   This is essential to run getKEGGModelFromHomology()
   when using a FASTA file as input
   WSL with a Linux distribution is not available, but HMMER
   requires it on Windows. Install it with "wsl --install" in
   a Command Prompt, restart, and rerun checkInstallation

=== Compatibility ===
 > Checking function uniqueness         Pass

*** checkInstallation complete ***

ans =

    'develop'

[edkerk]

You can ignore the HMM warning if you don't plan to run getKEGGModelFromHomology. I'll have the warning be more explicit about this.

If you do want to use getKEGGModelFromHomology we could explore further why it doesn't work.

[dimwas]

No, I do not want to use it, and if I will, I can run it on Ubuntu.
Thanks for solving my problem!

[edkerk]

Fix ended up in 2.11.4 release #760.