/Optimistic-Pessimistic-MetNet-KO

Primary LanguagePython

Metabolic Engineering Knockouts - Bilevel Linear Programming

You can use this package to solve the bilevel problem of reaction knockouts in metabolic engineering. I implemented an algorithm that exploits the bounds of the variables and adds feasibility cuts and no good cuts to find the best reaction knockout. The package includes data on E. Coli Bacteria, iJO1366, iAF1260, iJR904 and a Yeast Strain that were used to test the algorithm. But, the package also accepts custom bacteria.

Inputs

Parameteres for the algorithm use the params.json file.

  • pct_grow -> to set the growth percentage (minimal bacteria growth) to the wildtype "int" between 0-100.
  • K -> the number of allowed Knockouts "int".
  • Apr -> to distinguish between the optimistic and pessimistic aproach of the solution.
  • Bacteria -> to set the bacteria to use for the solution by wrting a text string with the names of the baceria. Otherwise, type in "Custom" for a custome bacteria and set the data in the "Custom_Inputs" here folder and custom chemical production.
  • FileName -> The name of the file where the results will be saved. The file is saved in the same folder where params.json and main.py are located.

Outputs

  • The name of the Bacteria.
  • The solving time from the solver in seconds.
  • The Strategy, that is the reaction Knockouts.
  • The chemical production for the specified chemical (objective function).
  • The biomass (cellular growth) for the mutant bacteria.
  • The binary vector that corresponds with the allowed reactions in the metabolic network (y=1) and the reaction knockouts (y=0).
  • The mass flow, that is the vector with the reactions' mass flow.

Run main.py on the terminal

  • P = Support.Parameters("params.json).assign_params() reads and sets the parameters from the json file
  • mn = Support.bacteriaselector(P.Bacteria) sets the bacteria
  • mn.tgt = P.pctgrow set the minimal growth in percentage to the maximum cellular grwoth for the wild type bacteria
  • solve = Support.Select_Approach(P.Apporach) sets the optimization approach
  • r = solve(network=mn,k=P.K,log=False) sets the solver parameters, log=True to print on the command window the optimization log
  • Support.SaveResults(result=r,filename=P.outfile,metnet=mn) saves the results to the specified out file in the json file

For a seamless work and no need for a major code modification keep the naming convention