dexom_python package top-level modules

Submodules

model_functions module

dexom_python.model_functions.check_constraint_values(model)

dexom_python.model_functions.check_model_options(model, timelimit=600, tolerance=1e-07, mipgaptol=0.001, verbosity=1)

Sets model options such as timelimit, tolerance, mipgapt tolerance, and verbosity

Parameters:

• model (cobra.Model) –

• timelimit (int) –

• tolerance (float) –

• mipgaptol (float) –

• verbosity (int) –

Returns:

model

Return type:

cobra.Model

dexom_python.model_functions.check_model_primals(model, rw, eps=0.01, thr=0.001, savename='primal_check.csv')

dexom_python.model_functions.check_threshold_tolerance(model, epsilon, threshold)

Checks if model.tolerance, epsilon and threshold parameters are compatible. If not, new epsilon and threshold values are returned.

Parameters:

• model (cobra.Model) –

• epsilon (float) –

• threshold (float) –

Returns:

• epsilon (float)

• threshold (float)

dexom_python.model_functions.get_all_reactions_from_model(model, save=True, shuffle=True, out_path='')

Outputs a list of all reactions in the model. If possible, all blocked reactions are removed. Optionally, the reaction-list can be shuffled.

Parameters:

• model (cobra.Model) –

• save (bool) – by default, exports the reactions in a csv format

• shuffle (bool) – set to True to shuffle the order of the reactions

• out_path (str) – output path

Returns:

rxn_list

Return type:

A list of all reactions in the model

dexom_python.model_functions.get_subsystems_from_model(model, save=True, out_path='')

Creates a list of all subsystems of a model and their associated reactions

Parameters:

• model (cobra.Model) –

• save (bool) –

• out_path (str) –

Returns:

• rxn_sub (pandas.DataFrame) – a DataFrame with reaction names as index and subsystem name as column

• sub_list (list) – a list of subsystems

dexom_python.model_functions.load_reaction_weights(filename, rxn_names='reactions', weight_names='weights')

loads reaction weights from a .csv file

Parameters:

• filename (str) – the path + name of a .csv file containing reaction weights

• rxn_names (str) – the name of the column containing the reaction names

• weight_names (str) – the name of the column containing the weights

Returns:

reaction_weights

Return type:

dict

dexom_python.model_functions.read_model(modelfile, solver='cplex')

Read a model from a file. Supported formats are SBML (.xml, .sbml), JSON (.json), and Matlab (.mat). Must use a cobrapy compatible solver.

Parameters:

• modelfile (str) –

• solver (str) –

Returns:

model

Return type:

cobra.Model

dexom_python.model_functions.save_reaction_weights(reaction_weights, filename='reaction_weights.csv')

Parameters:

• reaction_weights (dict) – a dictionary where keys = reaction IDs and values = weights

• filename (str) –

Returns:

reaction_weights

Return type:

pandas.DataFrame

gpr_rules module

dexom_python.gpr_rules.apply_gpr(model, gene_weights, save=True, filename='reaction_weights', duplicates='remove', null=0.0)

Applies the GPR rules from a given metabolic model for creating reaction weights

Parameters:

• model (cobra.Model) – a cobrapy model

• gene_weights (dict or pandas.Series or pandas.DataFrame) – a dictionary/pandas Series containing gene IDs as keys/index & weights as values

• save (bool) – if True, saves the reaction weights as a csv file

• filename (str) – path where the file will be saved

• duplicates (str, any of "remove", "max", "min", "mean", "median") – determines how to deal with genes presenting several expression values

• null (float) – value to return for reactions/genes with no information

Returns:

reaction_weights

Return type:

dict where keys = reaction IDs and values = weights

dexom_python.gpr_rules.expression2qualitative(genes, column_list=None, proportion=0.25, significant_genes='both', relative=True, save=True, outpath='geneweights')

Transforms gene expression values/ gene scores into qualitative gene weights

Parameters:

• genes (pandas.DataFrame or pandas.Series) – dataframe with gene IDs in the index and gene expression values in a later column

• column_list (list) – column indexes containing gene expression values to be transformed. If empty, all columns will be transformed

• proportion (tuple or float) – proportion of genes to be used for determining low and high gene expression

• significant_genes (str) – one of “high”, “low” or “both”. chooses whether the conversion is applied only for the genes with highest expression, lowest epxression, or both

• relative (bool) – if True, proportion parameter will be interpreted as a percentile if False, proportion parameter will be interpreted as an absolute value of gene expression

• save (bool) – if True, saves the resulting gene weights

• outpath (str) – if save=True, the .csv file will be saved to this path

Returns:

gene_weights – (-1 for low expression, 1 for high expression, 0 for in-between or no information)

Return type:

a pandas DataFrame containing qualitative gene weights

dexom_python.gpr_rules.replace_MulMax_AddMin(expression)

Function used for parsing gpr expressions

Parameters:

expression (symengine expression) – a symengine/sympy expression of a gpr rule

imat_functions module

exception dexom_python.imat_functions.ImatException

Bases: Exception

dexom_python.imat_functions.create_full_variable_single(model, rid, reaction_weights, epsilon, threshold)

dexom_python.imat_functions.create_full_variables(model, reaction_weights, epsilon, threshold)

Creates binary indicator variables in the model for every reaction.

dexom_python.imat_functions.create_new_partial_variable_single(model, rid, epsilon, threshold, pos)

dexom_python.imat_functions.create_new_partial_variables(model, reaction_weights, epsilon, threshold)

Creates binary indicator variables in the model for reactions with nonzero weight.

dexom_python.imat_functions.create_optimality_constraint(model, reaction_weights, prev_sol, obj_tol=0.001, name='optimality', full=False)

Creates the optimality constraint based on the imat objective function This constraint conserves the optimal objective value of the previous solution For detailed explanation of parameters see documentation of imat function.

Parameters:

• model (cobra.Model) –

• reaction_weights (dict) –

• prev_sol (cobra.Solution or float) – either the previous iMAT solution or the objective value of that solution

• obj_tol (float) – variance allowed in the objective_value of the solution

• name (string) –

• full (bool) –

Returns:

Return type:

optlang Constraint object (dependent on current solver)

dexom_python.imat_functions.imat(model, reaction_weights=None, epsilon=0.01, threshold=0.001, full=False)

Modified version of the integrative Metabolic Analysis Tool with reaction weights

Parameters:

• model (cobra.Model) – a cobrapy model

• reaction_weights (dict) – keys are reaction IDs, values are weights

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

• threshold (float) – activation threshold for all reactions

• full (bool) – if True, create variables for all reactions. if False, only for reactions with non-zero weights

Returns:

solution

Return type:

cobra.Solution

dexom_python.imat_functions.parsimonious_imat(model, reaction_weights=None, prev_sol=None, obj_tol=0.0, epsilon=0.01, threshold=0.001, full=False)

This function applies parsimonious iMAT. If no previous solution is supplied, an imat solution is computed. Then, the absoulte sum of reaction fluxes is minimzed, while maintaining the original objective value within a certain tolerance. For more complete description of parameters, see imat function.

Parameters:

• model (cobra.Model) –

• reaction_weights (dict) –

• prev_sol (cobra.Solution) – (optional) a previously computed imat solution

• obj_tol (float) – the relative tolerance on the objective function. By default, no tolerance is allowed

• epsilon (float) –

• threshold (float) –

• full (bool) –

Returns:

solution – a solution which maintains the imat objective while minimizing the absolute sum of fluxes

Return type:

cobra.Solution

result_functions module

dexom_python.result_functions.calc_objval_from_flux(fluxsol, model, rw, eps=0.01, thr=0.001)

Calculates the objective value of a solution based on the flux values

Parameters:

• fluxsol (pandas.Series) – a flux solution in which the index contains the model reaction IDs

• model (cobra.model) –

• rw (dict) –

• eps (float) –

• thr (float) –

Returns:

objval

Return type:

float

dexom_python.result_functions.compile_solutions(solutions, out_path='compiled_solutions', solution_pattern='*.csv', model=None, threshold=None)

Compiles individual solution files into one binary solution DataFrame

Parameters:

• solutions (list or str) – If list, must contain either solution files in .csv format, or Solution objects, or binary solution arrays. - If str, must be a folder in which the solution files math the sollution_pattern parameter

• out_path (str) – path to which the combined solutions will be saved

• solution_pattern (str) – If reading solutions from a folder, this is the pattern which will be used to recognize solution files

• model (cobrapy Model) – required if the solutions parameter is a list of Solution objects

• threshold (float) – required if the solutions parameter is a list of Solution objects

Returns:

sol_frame

Return type:

pandas DataFrame containg binary solutions

dexom_python.result_functions.plot_pca(solution_path, rxn_enum_solutions=None, save=True, save_name='')

Plots a 2-dimensional PCA of enumeration solutions

Parameters:

• solution_path (str) – csv file of enumeration solutions

• rxn_enum_solutions (str) – csv file of enumeration solutions. If specified, will plot these solutions in a different color

• save (bool) – if True, the pca-plot will be saved

• save_name (str) – name of the file to save

Returns:

pca

Return type:

sklearn.decomposition.PCA

dexom_python.result_functions.read_solution(filename, model=None, reaction_weights=None, solution_index=0, eps=0.01, thr=0.001)

Reads a solution from a .csv file. If the provided file is a file containing fluxes (as output by enumeration methods), the solution number solution_index will be read

Parameters:

• filename (str) – name of the file containing the solution

• model (cobra.Model) – optional unless the filename points to a solution file without reaction IDs

• reaction_weights (dict) – optional unless the filename points to a flux file, in which case it is needed for calculating objective values

• solution_index (int) – defines which solution will be read from the binary solution file

• eps (float) –

• thr (float) –

Returns:

• solution (cobra.Solution)

• sol_bin (numpy.array)

dexom_python.result_functions.write_solution(model, solution, threshold, filename='imat_sol.csv')

Writes an optimize solution as a txt file. The solution is written in a column format

Parameters:

• model (cobra.Model) –

• solution (cobra.Solution) –

• threshold (float) –

• filename (str) –

Returns:

• solution (cobra.Solution)

• solution_binary (numpy.array)

toy_models module

dexom_python.toy_models.create_reaction(model, rname, formula, gene_rule=None, fullname=None, lower_bound=0.0, upper_bound=1000.0, subsystem='')

dexom_python.toy_models.dagNet(num_layers, num_metabolites_per_layer, export=False, solver='cplex')

Creates a dagNet model where the metabolites of successive layers are all interconnected

Parameters:

• num_layers (int) – number of layers

• num_metabolites_per_layer (int) – number of metabolites per layer

• export (bool) – if True, exports the model as .json and the reaction weights as .csv

• solver (str) – a valid cobrapy solver

Returns:

• model (cobra.Model)

• reaction_weights (dict)

dexom_python.toy_models.r13m10(export=False, solver='cplex')

This is the model that is used for the unit tests in the tests/model folder

Parameters:

• export (bool) – if True, exports the model as .json, .xml and .mat, and the reaction weights as .csv

• solver (str) – a valid cobrapy solver

Returns:

• model (cobra.Model)

• reaction_weights (dict)

dexom_python.toy_models.small4M(export=False, solver='cplex')

creates the small4M model

Parameters:

• export (bool) – if True, exports the model as .json and the reaction weights as .csv

• solver (str) – a valid cobrapy solver

Returns:

• model (cobra.Model)

• reaction_weights (dict)

dexom_python.toy_models.small4S(export=False, solver='cplex')

creates the small4S model

Parameters:

• export (bool) – if True, exports the model as .json and the reaction weights as .csv

• solver (str) – a valid cobrapy solver

Returns:

• model (cobra.Model)

• reaction_weights (dict)

pathway_enrichment module

dexom_python.pathway_enrichment.Fisher_groups(model, solpath, outpath='Fisher_groups')

!!! This only works if the pathway name is stored in the model.groups property !!! For models where the pathways are stored in the model.reactions.subsystem property, use the Fischer_subsystems function

Performs pathway over- and underrepresentation analysis

Parameters:

• model (cobra.Model) –

• solpath (str) – file containing DEXOM solutions

• outpath (str) – path to which results are saved

Returns:

over, under

Return type:

pandas.DataFrames (saved as .csv files) containing -log10 BH-adjusted p-values

dexom_python.pathway_enrichment.Fisher_subsystems(solpath, subframe, sublist, outpath='Fisher_subsystems')

!!! This only works if the pathway name is stored in the model.reaction.subsystem property !!! For models where the pathways are stored in the model.groups property, use the Fischer_groups function

Performs pathway over- and underrepresentation analysis

Parameters:

• solpath (str) – file containing DEXOM solutions

• subframe (pandas.DataFrame) – dataframe with ‘ID’ column containing reaction IDs and one ‘subsystem’ column containing pathways

• sublist (list) – list of subsystems

• outpath (str) – path to which results are saved

Returns:

over, under

Return type:

pandas.DataFrames (saved as .csv files) containing -log10 BH-adjusted p-values

dexom_python.pathway_enrichment.perform_ora(gene_list, pathway_genes_dict, total_genes)

Smaller function for performing ORA without reading files

Parameters:

• gene_list (list) – list of genes

• pathway_genes_dict (dict) – dictionary with pathway names as keys and list of genes as values

• total_genes (int) – total number of genes in the background set

Returns:

pandas.Series

Return type:

sorted p-values with BH-correction

dexom_python.pathway_enrichment.plot_Fisher_pathways(filename_over, filename_under, sublist, outpath='pathway_enrichment')

Module contents