dexom_python.enum_functions enumeration modules

Submodules

enumeration module

class dexom_python.enum_functions.enumeration.EnumSolution(solutions, binary, objective_value)

Bases: object

class for solutions of enumeration methods

Parameters

• solutions (list) – A list of pandas dataframes containing flux values with reaction ids as index

• binary (list) – A list containing binary arrays of reaction activity (0 for inactive, 1 for active)

• objective_value (float) – objective value returned by the solver at the end of the optimization

dexom_python.enum_functions.enumeration.analyze_div_enum_results(result_path, solution_path, out_path)

This function calculates the average pairwise hamming distance and average next neighbour distance for each iteration - it’s very slow

Parameters

• result_path (csv results file from diversity-enum) –

• solution_path (csv solution file from diversity-enum) –

• out_path (path for saving) –

dexom_python.enum_functions.enumeration.create_enum_variables(model, reaction_weights, eps=0.01, thr=1e-05, full=False)

dexom_python.enum_functions.enumeration.get_recent_solution_and_iteration(dirpath, startsol_num)

This functions fetches a solution from a given directory. The solutions are ordered by creation time, and one solution is picked using an exponential distribution (meaning that the most recent solution has the highest probability of being chosen)

Parameters

• dirpath (str) – a directory containing imat or enumeration solutions

• startsol_num (int) – the number of starting solutions present in the directory

Returns

• solution (a Solution object)

• iteration (int) – calculates the current iteration, based on how many solutions are already present in the folder

dexom_python.enum_functions.enumeration.write_batch_script1(directory, modelfile, weightfile, cplexpath, reactionlist=None, imatsol=None, objtol=0.01, filenums=100, iters=100)

Writes bash scripts for dexom-python parallelization approach 1 on a slurm cluster. Within each batch, reaction-enumeration and diversity-enumeration are performed. These scripts assume that you have setup a virtual environment called env.

Parameters

• directory (str) – directory in which the files will be generated. If it does not exist, it will be created

• modelfile (str) – path to the model

• weightfile – path to the reaction weights

• cplexpath (str) – path to a cplex installation on the cluster

• reactionlist (str) – list of reactions for reaction-enumeration

• imatsol (str) – path to imat solution

• objtol (float) – objective tolerance

• filenums (int) – number of parallel batches

• iters (int) – number of diversity-enumeration iterations per batch

dexom_python.enum_functions.enumeration.write_batch_script2(directory, modelfile, weightfile, cplexpath, objtol=0.01, filenums=100)

Writes bash scripts for dexom-python parallelization approach 2 on a slurm cluster. In this approach, indiviual diversity-enumeration iterations are laucnhed in each batch - this requires the existance of reaction-enumeration solutions beforehand. These scripts assume that you have setup a virtual environment called env.

Parameters

• directory (str) – directory in which the files will be generated

• modelfile (str) – path to the model

• weightfile – path to the reaction weights

• cplexpath (str) – path to a cplex installation on the cluster

• objtol (float) – objective tolerance

• filenums (int) – number of parallel batches

dexom_python.enum_functions.enumeration.write_batch_script_divenum(directory, modelfile, weightfile, rxnsols, objtol, eps=0.0001, thr=1e-05, tol=1e-08, filenums=100, iters=100, t=6000)

dexom_python.enum_functions.enumeration.write_rxn_enum_script(directory, modelfile, weightfile, imatsol=None, reactionlist=None, eps=0.0001, thr=1e-05, tol=1e-08, iters=100, maxiters=10000000000.0)

rxn_enum_functions module

class dexom_python.enum_functions.rxn_enum_functions.RxnEnumSolution(all_solutions, unique_solutions, all_binary, unique_binary, all_reactions=None, unique_reactions=None, objective_value=-1.0)

Bases: object

dexom_python.enum_functions.rxn_enum_functions.rxn_enum(model, reaction_weights, prev_sol=None, rxn_list=[], eps=0.0001, thr=0.0001, obj_tol=0.001, out_path='enum_rxn', save=False)

Reaction enumeration method

Parameters

• model (cobrapy Model) –

• reaction_weights (dict) – keys = reactions and values = weights

• prev_sol (imat Solution object) – an imat solution used as a starting point

• eps (float) – activation threshold in imat

• thr (float) – detection threshold of activated reactions

• tlim (int) – time limit for imat

• tol (float) – tolerance for imat

• obj_tol (float) – variance allowed in the objective_values of the solutions

• out_path (str) – path to which the solutions are saved if save==True

• save (bool) – if True, every individual solution is saved in the iMAT solution format

Returns

solution

Return type

RxnEnumSolution object

icut_functions module

dexom_python.enum_functions.icut_functions.create_icut_constraint(model, reaction_weights, threshold, prev_sol, name, full=False)

Creates an icut constraint on the previously found solution. This solution is excluded from the solution space.

dexom_python.enum_functions.icut_functions.icut(model, reaction_weights, prev_sol=None, eps=0.0001, thr=0.0001, obj_tol=0.001, maxiter=10, full=False)

integer-cut method

Parameters

• model (cobrapy Model) –

• reaction_weights (dict) – keys = reactions and values = weights

• prev_sol (imat Solution object) – an imat solution used as a starting point

• eps (float) – activation threshold in imat

• thr (float) – detection threshold of activated reactions

• obj_tol (float) – variance allowed in the objective_values of the solutions

• maxiter (foat) – maximum number of solutions to check for

• full (bool) – if True, carries out integer-cut on all reactions; if False, only on reactions with non-zero weights

Returns

solution – In the case of integer-cut, all_solutions and unique_solutions are identical

Return type

EnumSolution object

maxdist_functions module

dexom_python.enum_functions.maxdist_functions.create_maxdist_constraint(model, reaction_weights, prev_sol, obj_tol, name='maxdist_optimality', full=False)

Creates the optimality constraint for the maxdist algorithm. This constraint conserves the optimal objective value of the previous solution

dexom_python.enum_functions.maxdist_functions.create_maxdist_objective(model, reaction_weights, prev_sol, prev_sol_bin, only_ones=False, full=False)

Create the new objective for the maxdist algorithm. This objective is the minimization of similarity between the binary solution vectors If only_ones is set to False, the similarity will only be calculated with overlapping ones

dexom_python.enum_functions.maxdist_functions.maxdist(model, reaction_weights, prev_sol=None, eps=0.0001, thr=0.0001, obj_tol=0.001, maxiter=10, icut=True, full=False, only_ones=False)

Parameters

• model (cobrapy Model) –

• reaction_weights (dict) – keys = reactions and values = weights

• prev_sol (imat Solution object) – an imat solution used as a starting point

• eps (float) – activation threshold in imat

• thr (float) – detection threshold of activated reactions

• obj_tol (float) – variance allowed in the objective_values of the solutions

• maxiter (foat) – maximum number of solutions to check for

• icut (bool) – if True, icut constraints are applied

• full (bool) – if True, carries out integer-cut on all reactions; if False, only on reactions with non-zero weights

• only_ones (bool) – if True, only the ones in the binary solution are used for distance calculation (as in dexom matlab)

Returns

solution

Return type

EnumSolution object

diversity_enum_functions module

dexom_python.enum_functions.diversity_enum_functions.diversity_enum(model, reaction_weights, prev_sol=None, eps=0.0001, thr=0.0001, obj_tol=0.001, maxiter=10, dist_anneal=0.995, out_path='enum_dexom', icut=True, full=False, save=False)

diversity-based enumeration

Parameters

• model (cobrapy Model) –

• reaction_weights (dict) – keys = reactions and values = weights

• prev_sol (Solution instance) – a previous imat solution

• eps (float) – activation threshold for highly expressed reactions

• thr (float) – detection threshold of activated reactions

• obj_tol (float) – variance allowed in the objective_values of the solutions

• maxiter (foat) – maximum number of solutions to search for

• dist_anneal (float) – parameter which influences the probability of selecting reactions

• out_path (str) – path to which the solutions are saved if save==True

• icut (bool) – if True, icut constraints are applied

• full (bool) – if True, the full-DEXOM implementation is used

• save (bool) – if True, every individual solution is saved in the iMAT solution format

Returns

solution

Return type

an EnumSolution object

Module contents