Welcome to the documentation of PyTFA !¶

TFA is a method that builds on FBA to improve its solution space. Specifically, it includes thermodynamics and explicit formulation of Gibbs energies and metabolite concentrations, which enables straightforward integration of metabolite concentration measurements.

If you use our work, please cite us 1 !

1: Salvy, P., Fengos, G., Ataman, M., Pathier, T., Soh, K. C., & Hatzimanikatis, V. (2018). pyTFA and matTFA: a Python package and a Matlab toolbox for Thermodynamics-based Flux Analysis. Bioinformatics, 35(1), 167-169.

Quick start¶

Three tutorial files detail thoroughly normal usages of the pytfa package. They can be found at:

pytfa
└── tutorials
    ├── figure_paper.py
    ├── tutorial_basics.py
    └── tutorial_sampling.py

figure_paper.py details how to get the figure from our paper 1, a simple use case for TFA on a reduced Escherichia coli. We show that adding thermodyamics constraints and simple concentration data allow to substantially reduce the flux space.

tutorial_basics.py shows a more realistic case with two models (reduced or full genome-scale) of Escherichia coli. It also cycles through several solvers (if more are installed), to show how simple it is to change your solver (thanks to optlang).

tutorial_sampling.py shows how to sample a variable, for example thermodynamic displacement, and generate plots to visualize the results.

If you plan to run the tutorials with full genome-scale models, we recommend you to get a commercial solver, as it has been seen that GLPK’s lack of parallelism significantly increases solving time

The next sections give more details on how the thermodynamic model is structured, and how data is managed.

Cheers,

The py.TFA team

1: Salvy, P., Fengos, G., Ataman, M., Pathier, T., Soh, K. C., & Hatzimanikatis, V. (2018). pyTFA and matTFA: a Python package and a Matlab toolbox for Thermodynamics-based Flux Analysis. Bioinformatics, 35(1), 167-169.

pyTFA models¶

pyTFA models are based on COBRApy models, with additional values.

Compartment data¶

This is the compartments attribute of the model. It is a dict where each key is the symbol of a compartment, and the value is another dict with the following keys :

c_min	`float` The minimum concentration for each metabolite in the compartment, in mol.L-1
c_max	`float` The maximum concentration for each metabolite in the compartment, in mol.L-1
ionicStr	`float` The ionic strength of the compartment (mV)
membranePot	`dict` A dictionnary representing the membrane potential between this compartment (which is the source compartment) and the others. Each key is the symbol of another compartment (which is the destination compartment), and the value is the potential (in mV) from the source to the destination.
name	`string` The name of the compartment
pH	`float` The pH in the compartment
symbol	`string` The symbol of the compartment (which is the key of this dictionnary)

Here is an example:

cobra_model.compartments['c'] = {
  'c_max': 0.01,
  'c_min': 9.9999999999999995e-08,
  'ionicStr': 0.25,
  'membranePot': {
    'c': 0,
    'e': 60,
    'g': 0,
    'm': -180,
    'n': 0,
    'p': 0,
    'r': 0,
    't': 0,
    'v': 0,
    'x': 0
  },
  'name': 'Cytosol',
  'pH': 7.0,
  'symbol': 'c'
}

Metabolites¶

Each metabolite must be annotated with its SeedID, which will be used to get the thermodynamic values from the Thermodynamic Databases. In order to do this, use the annotation attribute of each metabolite. Here is an example:

cobra_model.metabolites[0].annotation = {
  'SeedID': 'cpd00018'
}

pyTFA will also define a thermo a thermo attribute for each metabolite, which is a pytfa.thermo.MetaboliteThermo.

Reactions¶

pyTFA will define a thermo attribute for each reaction. It is a dict with the following attributes:

computed	`bool` Whether the thermodynamic values were computed or not.
deltaGR	`float` Sum of the non-concentration terms for the reaction. Used as the right-hand side value of a constraint. If the thermodynamic values were not computed, this is `10**7`.
deltaGRerr	`float` Error on `deltaGR` If the thermodynamic values were not computed, this is `10**7`.
deltaGrxn	`float` Sum of the deltaGF of all the metabolites in the reaction. Not defined if computed is False !
isTrans	`bool` Whether the reaction is a transport reaction or not

Here are some examples:

cobra_model.reactions[0].thermo = {
  'computed': False,
  'deltaGR': 10000000,
  'deltaGRerr': 10000000,
  'isTrans': False
}

cobra_model.reactions[99].thermo = {
  'computed': True,
  'deltaGR': 1.161097833014658,
  'deltaGRerr': 2,
  'deltaGrxn': 0,
  'isTrans': True,
}

Thermodynamic Databases¶

Converting a Matlab thermodynamic database¶

If you have a Matlab thermodynamic database, you can easily convert it to a Python database thanks to the script thermoDBconverter.py:

python thermoDBconverter.py database.mat converted_database.thermodb

Loading a thermodynamic database¶

Thermodynamic databases are compressed through zlib and binary-encoded with pickle. In order to load them, you need first to uncompress them with zlib.decompress then load the result into memory with pickle.loads:

import pickle
import zlib
with open('thermoDatabases/DB_AlbertyUpdate.thermodb', 'rb') as file:
    ReactionDB = pickle.loads(zlib.decompress(file.read()))

Warning

Since the file is compressed, you MUST load it as a binary file by calling open with the b flag, otherwise Python will try to decode it as unicode and raise an exception !

Structure of a thermodynamic database¶

A thermodynamic database is a dict with the following fields:

name : string The name of the database

units : string The unit of the energies in the database. Can be kcal/mol or kJ/mol.

metabolites : dict A dictionnary containing the metabolites’ thermodynamic data. See Metabolites for more information.

cues : dict A dictionnary containing the cues’ thermodynamic data. See Cues for more information.

Metabolites¶

This is a dictionnary storing various thermodynamic data about metabolites. It is stored as a dict where each key is a SeedID. The values are others dict with the following keys.

id	`string` SeedID of the metabolite
charge_std	`float` Charge of the metabolite (mV) in standard conditions
deltaGf_std	`float` Transformed Gibbs energy of formation of the metabolite, in standard conditions.
deltaGf_err	`float` Error on the transformed Gibbs energy of formation of the metabolite, in standard conditions
mass_std	`float` Mass of the metabolite (g.mol-1)
nH_std	`int` Number of protons of the metabolite, in standard conditions
error	`string` Error on the metabolite’s thermodynamic data. Thermodynamic values will be computed only if this equals to ‘Nil’.
formula	`string` Formula of the metabolite.
nH_std	`int` Number of protons in the metabolite’s formula
name	`string` Name of the metabolite
other_names	`list` (`string`) Other names of the metabolite
pKa	`list` (`float`) pKas of the metabolite
struct_cues	`dict` (`int`) cues of the metabolite The keys of the array are the names of the cues, and the values the number of cues of this type that are part of the structure.

Here is an example:

ReactionDB['metabolites']['cpd00001'] = {
  'charge_std': 0,
  'deltaGf_err': 0.5,
  'deltaGf_std': -56.686999999999998,
  'error': 'Nil',
  'formula': 'H2O',
  'id': 'cpd00001',
  'mass_std': 18.0,
  'nH_std': 2,
  'name': 'H2O',
  'other_names': ['H2O', 'Water', 'HO-', 'OH-', 'h2o'],
  'pKa': [15.7],
  'struct_cues': {'H2O': 1}
}

Cues¶

This is a dictionnary storing various thermodynamic data about cues. It is stored as a dict where each key is the cue ID, as referrenced in the struct_cues attribute of Metabolites. The values are others dict with the following keys.

id	`string` ID of the cue
charge	`float` The charge (mV) of the cue in standard conditions
datfile	`string` The dat file from which the data was imported. Optional
energy	`float` Transformed Gibbs energy of formation of the cue, in standard conditions.
error	`float` The error on the transformed Gibbs energy of formation of the cue, in standard conditions.
formula	`string` Formula of the cue
names	`list` (`string`) Other names of the cue
small	`bool` Whether this is a small cue or not

Here is an example:

ReactionDB['cues']['H2O'] = {
  'charge': 0,
  'datfile': 'H2O.gds',
  'energy': -56.686999999999998,
  'error': 0.5,
  'formula': 'H2O',
  'id': 'H2O',
  'names': ['H2O', 'OH-', 'HO-'],
  'small': True
}

Solver setup¶

This document is written assuming a Docker container installation. However, you can easily adapt the content to other types of Linux-based installations.

GPLK¶

GLPK should be directly available from the requirements.

CPLEX¶

You will need to first install CPLEX on a Linux machine.

Place in etfl/docker/solvers/ the folder /ibm that is installed by CPLEX (usually in /opt). You actually only need the following substructure (makes the container lighter):

.
└───ibm
    └───ILOG
        └───CPLEX_StudioXXXX
            └───cplex
                ├───bin
                ├───include
                ├───lib
                └───python

Gurobi¶

Place in etfl/docker/solvers/ the tarball you downloaded from the website, and modify accordingly the files:

../utils/install_gurobi.sh
../utils/activate_gurobi.sh

Make sure you change the paths and filenames to reflect the actual version of Gurobi you are running.

Gurobi needs a floating license for Docker instances, (see http://www.gurobi.com/documentation/7.5/quickstart_windows/setting_up_and_using_a_flo.html#subsection:tokenserver) Once your system administrator set it up, you will need to add your gurobi license server to ../utils/gurobi.lic.template, and rename it to gurobi.lic

Integrating metabolomics¶

In this short example we will go through a simple case of integration of absolute metabolite concentrations.

Let us imagine we got absolute concentration values for cytosolic ATP:

\[5 . 10^{-3} \; mol.L^{-1} \le [X] \le 3 . 10^{-2} \; mol.L^{-1}\]

Then you can tell the model that your (log) concentration is limited in range:

from math import log

mymodel.log_concentration.atp_c.variable.lb = log(5e-3)
mymodel.log_concentration.atp_c.variable.ub = log(3e-2)

This will constrain the dG according to your concentration measurements for cytosolic ATP. As a reminder, the dG (not the dGo) takes activity (here, concentrations) into account for its calculation. You can find a more detailed explanation in those papers:

Henry, Christopher S., Linda J. Broadbelt, and Vassily Hatzimanikatis. “Thermodynamics-based metabolic flux analysis.” Biophysical journal 92.5 (2007): 1792-1805.
Soh, Keng Cher, Ljubisa Miskovic, and Vassily Hatzimanikatis. “From network models to network responses: integration of thermodynamic and kinetic properties of yeast genome-scale metabolic networks.” FEMS yeast research 12.2 (2012): 129-143.

API Reference¶

This page contains auto-generated API reference documentation 1.

`pytfa`¶

Thermodynamic analysis for Flux-Based Analysis

Subpackages¶

`pytfa.analysis`¶

Submodules¶

pytfa.analysis.chebyshev¶

Variability analysis

Module Contents¶

Classes¶

ChebyshevRadius

Variable representing a Chebyshev radius

Functions¶

`is_inequality`(constraint)
`chebyshev_center`(model, variables, inplace=False, big_m=BIGM, include=list(), exclude=list())	Computes the chebyshev center of a problem with respect to given variables,
`chebyshev_transform`(model, vars, include_list=list(), exclude_list=list(), radius_id='radius', scaling_factor=1, big_m=BIGM)	Adds a Chebyshev radius variable and edits accordingly the selected
`get_cons_var_classes`(model, elements, type)
`get_variables`(model, variables)

Attributes¶

BIGM

pytfa.BIGM = 1000¶

class pytfa.ChebyshevRadius(model, id_, **kwargs)¶

Bases: pytfa.optim.variables.ModelVariable

Variable representing a Chebyshev radius

prefix = CR_¶

pytfa.is_inequality(constraint)¶

pytfa.chebyshev_center(model, variables, inplace=False, big_m=BIGM, include=list(), exclude=list())¶

Computes the chebyshev center of a problem with respect to given variables, including `include’ constraints and excluding `exclude’ constraints. Warning: Only works with pyTFA variables so far

Parameters

model –
variables –
inplace –
big_m –

Returns

pytfa.chebyshev_transform(model, vars, include_list=list(), exclude_list=list(), radius_id='radius', scaling_factor=1, big_m=BIGM)¶

Adds a Chebyshev radius variable and edits accordingly the selected constraints

Parameters

model –
vars – variables with respect to which to perform the Chebyshev centering. If none is supplied, all of the variables in the equation will be considered
include_list –
exclude_list –
radius_id –
big_m –

Returns

pytfa.get_cons_var_classes(model, elements, type)¶

pytfa.get_variables(model, variables)¶

pytfa.analysis.manipulation¶

Module Contents¶

Functions¶

`apply_reaction_variability`(tmodel, va, inplace=True)	Applies the VA results as bounds for the reactions of a cobra_model
`apply_generic_variability`(tmodel, va, inplace=True)	Reactions a dealt with cobra, but the other variables added use pytfa's
`apply_directionality`(tmodel, solution, inplace=True)	Takes a flux solution and transfers its reaction directionality as

pytfa.analysis.manipulation.apply_reaction_variability(tmodel, va, inplace=True)[source]¶: Applies the VA results as bounds for the reactions of a cobra_model :param inplace: :param tmodel: :param va: :return:

pytfa.analysis.manipulation.apply_generic_variability(tmodel, va, inplace=True)[source]¶

Reactions a dealt with cobra, but the other variables added use pytfa’s interface: the class GenericVariable. We use a different method to apply variability directly in the solver

Parameters

tmodel –
va –
inplace –

Returns

pytfa.analysis.manipulation.apply_directionality(tmodel, solution, inplace=True)[source]¶

Takes a flux solution and transfers its reaction directionality as constraints for the cobra_model

Parameters

inplace –
tmodel –
solution –

Returns

pytfa.analysis.sampling¶

Sampling wrappers for pytfa models

Module Contents¶

Classes¶

`GeneralizedHRSampler`	The abstract base class for hit-and-run samplers.
`GeneralizedACHRSampler`	The abstract base class for hit-and-run samplers.
`GeneralizedOptGPSampler`	The abstract base class for hit-and-run samplers.

Functions¶

sample(model, n, method='optgp', thinning=100, processes=1, seed=None)

Sample valid flux distributions from a thermo cobra_model.

class pytfa.GeneralizedHRSampler(model, thinning, nproj=None, seed=None)¶

Bases: cobra.sampling.HRSampler

The abstract base class for hit-and-run samplers.

New samplers should derive from this class where possible to provide a uniform interface.

modelcobra.Model: The cobra model from which to generate samples.
thinningint: The thinning factor of the generated sampling chain. A thinning of 10 means samples are returned every 10 steps.
nprojint > 0, optional: How often to reproject the sampling point into the feasibility space. Avoids numerical issues at the cost of lower sampling. If you observe many equality constraint violations with sampler.validate you should lower this number (default None).
seedint > 0, optional: Sets the random number seed. Initialized to the current time stamp if None (default None).

feasibility_tol: float: The tolerance used for checking equalities feasibility.
bounds_tol: float: The tolerance used for checking bounds feasibility.
n_samplesint: The total number of samples that have been generated by this sampler instance.
retriesint: The overall of sampling retries the sampler has observed. Larger values indicate numerical instabilities.
problemProblem: A NamedTuple whose attributes define the entire sampling problem in matrix form.
warmupnumpy.matrix: A numpy matrix with as many columns as reactions in the model and more than 3 rows containing a warmup sample in each row. None if no warmup points have been generated yet.
fwd_idxnumpy.array: A numpy array having one entry for each reaction in the model, containing the index of the respective forward variable.
rev_idxnumpy.array: A numpy array having one entry for each reaction in the model, containing the index of the respective reverse variable.

generate_fva_warmup(self)¶

Generate the warmup points for the sampler.

Generates warmup points by setting each flux as the sole objective and minimizing/maximizing it. Also caches the projection of the warmup points into the nullspace for non-homogeneous problems (only if necessary).

class pytfa.GeneralizedACHRSampler(model, thinning=100, seed=None)¶

Bases: GeneralizedHRSampler, cobra.sampling.ACHRSampler

The abstract base class for hit-and-run samplers.

New samplers should derive from this class where possible to provide a uniform interface.

modelcobra.Model: The cobra model from which to generate samples.
thinningint: The thinning factor of the generated sampling chain. A thinning of 10 means samples are returned every 10 steps.
nprojint > 0, optional: How often to reproject the sampling point into the feasibility space. Avoids numerical issues at the cost of lower sampling. If you observe many equality constraint violations with sampler.validate you should lower this number (default None).
seedint > 0, optional: Sets the random number seed. Initialized to the current time stamp if None (default None).

feasibility_tol: float: The tolerance used for checking equalities feasibility.
bounds_tol: float: The tolerance used for checking bounds feasibility.
n_samplesint: The total number of samples that have been generated by this sampler instance.
retriesint: The overall of sampling retries the sampler has observed. Larger values indicate numerical instabilities.
problemProblem: A NamedTuple whose attributes define the entire sampling problem in matrix form.
warmupnumpy.matrix: A numpy matrix with as many columns as reactions in the model and more than 3 rows containing a warmup sample in each row. None if no warmup points have been generated yet.
fwd_idxnumpy.array: A numpy array having one entry for each reaction in the model, containing the index of the respective forward variable.
rev_idxnumpy.array: A numpy array having one entry for each reaction in the model, containing the index of the respective reverse variable.

class pytfa.GeneralizedOptGPSampler(model, processes, thinning=100, seed=None)¶

Bases: GeneralizedHRSampler, cobra.sampling.OptGPSampler

The abstract base class for hit-and-run samplers.

New samplers should derive from this class where possible to provide a uniform interface.

modelcobra.Model: The cobra model from which to generate samples.
thinningint: The thinning factor of the generated sampling chain. A thinning of 10 means samples are returned every 10 steps.
nprojint > 0, optional: How often to reproject the sampling point into the feasibility space. Avoids numerical issues at the cost of lower sampling. If you observe many equality constraint violations with sampler.validate you should lower this number (default None).
seedint > 0, optional: Sets the random number seed. Initialized to the current time stamp if None (default None).

feasibility_tol: float: The tolerance used for checking equalities feasibility.
bounds_tol: float: The tolerance used for checking bounds feasibility.
n_samplesint: The total number of samples that have been generated by this sampler instance.
retriesint: The overall of sampling retries the sampler has observed. Larger values indicate numerical instabilities.
problemProblem: A NamedTuple whose attributes define the entire sampling problem in matrix form.
warmupnumpy.matrix: A numpy matrix with as many columns as reactions in the model and more than 3 rows containing a warmup sample in each row. None if no warmup points have been generated yet.
fwd_idxnumpy.array: A numpy array having one entry for each reaction in the model, containing the index of the respective forward variable.
rev_idxnumpy.array: A numpy array having one entry for each reaction in the model, containing the index of the respective reverse variable.

pytfa.sample(model, n, method='optgp', thinning=100, processes=1, seed=None)¶

Sample valid flux distributions from a thermo cobra_model.

Function adapted from cobra.flux_analysis.sample to display all solver variables

Documentation adapted from cobra.flux_analysis.sample

‘optgp’ (default) which uses the OptGPSampler that supports parallel
analysis 1. Requires large numbers of samples to be performant (n < 1000). For smaller samples ‘achr’ might be better suited.

or

‘achr’ which uses artificial centering hit-and-run. This is a single process method with good convergence 2.

modelpytfa.core.ThermoModel: The cobra_model from which to sample variables.
nint: The number of samples to obtain. When using ‘optgp’ this must be a multiple of processes, otherwise a larger number of samples will be returned.
methodstr, optional: The analysis algorithm to use.
thinningint, optional: The thinning factor of the generated analysis chain. A thinning of 10 means samples are returned every 10 steps. Defaults to 100 which in benchmarks gives approximately uncorrelated samples. If set to one will return all iterates.
processesint, optional: Only used for ‘optgp’. The number of processes used to generate samples.
seedpositive integer, optional: The random number seed to be used. Initialized to current time stamp if None.

pandas.DataFrame: The generated flux samples. Each row corresponds to a sample of the fluxes and the columns are the reactions.

The samplers have a correction method to ensure equality feasibility for long-running chains, however this will only work for homogeneous models, meaning models with no non-zero fixed variables or constraints ( right-hand side of the equalities are zero).

1: Megchelenbrink W, Huynen M, Marchiori E (2014) optGpSampler: An Improved Tool for Uniformly Sampling the Solution-Space of Genome-Scale Metabolic Networks. PLoS ONE 9(2): e86587.
2: Direction Choice for Accelerated Convergence in Hit-and-Run Sampling David E. Kaufman Robert L. Smith Operations Research 199846:1 , 84-95

pytfa.analysis.variability¶

Variability analysis

Module Contents¶

Functions¶

`find_bidirectional_reactions`(va, tolerance=1e-08)	Returns the ids of reactions that can both carry net flux in the forward or
`find_directionality_profiles`(tmodel, bidirectional, max_iter=10000.0, solver='optlang-glpk', tolerance=1e-09)	Takes a ThermoModel and performs enumeration of the directionality profiles
`_bool2str`(bool_list)	turns a list of booleans into a string
`_variability_analysis_element`(tmodel, var, sense)
`variability_analysis`(tmodel, kind='reactions', proc_num=BEST_THREAD_RATIO)	Performs variability analysis, gicven a variable type
`parallel_variability_analysis`(tmodel, kind='reactions', proc_num=BEST_THREAD_RATIO)	WIP.
`calculate_dissipation`(tmodel, solution=None)

Attributes¶

`CPU_COUNT`
`BEST_THREAD_RATIO`

pytfa.CPU_COUNT¶

pytfa.BEST_THREAD_RATIO¶

pytfa.find_bidirectional_reactions(va, tolerance=1e-08)¶

Returns the ids of reactions that can both carry net flux in the forward or backward direction.

Parameters

va –

A variability analysis, pandas Dataframe like so:: maximum minimum

6PGLter -8.330667e-04 -8.330667e-04 ABUTt2r 0.000000e+00 0.000000e+00 ACALDt 0.000000e+00 0.000000e+00

Returns

pytfa.find_directionality_profiles(tmodel, bidirectional, max_iter=10000.0, solver='optlang-glpk', tolerance=1e-09)¶

Takes a ThermoModel and performs enumeration of the directionality profiles

Parameters

tmodel –
max_iter –

Returns

pytfa._bool2str(bool_list)¶

turns a list of booleans into a string

Parameters: bool_list – ex: ‘[False True False False True]’
Returns: ‘01001’

pytfa._variability_analysis_element(tmodel, var, sense)¶

pytfa.variability_analysis(tmodel, kind='reactions', proc_num=BEST_THREAD_RATIO)¶

Performs variability analysis, gicven a variable type

Parameters

tmodel –
kind –
proc_num –

Returns

pytfa.parallel_variability_analysis(tmodel, kind='reactions', proc_num=BEST_THREAD_RATIO)¶

WIP.

Parameters

tmodel –
kind –
proc_num –

Returns

pytfa.calculate_dissipation(tmodel, solution=None)¶

Package Contents¶

Classes¶

`GeneralizedHRSampler`	The abstract base class for hit-and-run samplers.
`GeneralizedACHRSampler`	The abstract base class for hit-and-run samplers.
`GeneralizedOptGPSampler`	The abstract base class for hit-and-run samplers.
`DeltaG`	Class to represent a DeltaG
`ForbiddenProfile`	Class to represent a forbidden net flux directionality profile
`ForwardUseVariable`	Class to represent a forward use variable, a type of binary variable used to

Functions¶

`sample`(model, n, method='optgp', thinning=100, processes=1, seed=None)	Sample valid flux distributions from a thermo cobra_model.
`get_direction_use_variables`(tmodel, solution)	Returns the active use variables in a solution. Use Variables are binary
`get_active_use_variables`(tmodel, solution)	Returns the active use variables in a solution. Use Variables are binary
`get_bistream_logger`(name)	Sets up a logger that outputs INFO+ messages on stdout and DEBUG+ messages
`find_bidirectional_reactions`(va, tolerance=1e-08)	Returns the ids of reactions that can both carry net flux in the forward or
`find_directionality_profiles`(tmodel, bidirectional, max_iter=10000.0, solver='optlang-glpk', tolerance=1e-09)	Takes a ThermoModel and performs enumeration of the directionality profiles
`_bool2str`(bool_list)	turns a list of booleans into a string
`_variability_analysis_element`(tmodel, var, sense)
`variability_analysis`(tmodel, kind='reactions', proc_num=BEST_THREAD_RATIO)	Performs variability analysis, gicven a variable type
`parallel_variability_analysis`(tmodel, kind='reactions', proc_num=BEST_THREAD_RATIO)	WIP.
`calculate_dissipation`(tmodel, solution=None)
`apply_reaction_variability`(tmodel, va, inplace=True)	Applies the VA results as bounds for the reactions of a cobra_model
`apply_generic_variability`(tmodel, va, inplace=True)	Reactions a dealt with cobra, but the other variables added use pytfa's
`apply_directionality`(tmodel, solution, inplace=True)	Takes a flux solution and transfers its reaction directionality as

Attributes¶

`CPU_COUNT`
`BEST_THREAD_RATIO`

class pytfa.analysis.GeneralizedHRSampler(model, thinning, nproj=None, seed=None)[source]¶

Bases: cobra.sampling.HRSampler

The abstract base class for hit-and-run samplers.

New samplers should derive from this class where possible to provide a uniform interface.

modelcobra.Model: The cobra model from which to generate samples.
thinningint: The thinning factor of the generated sampling chain. A thinning of 10 means samples are returned every 10 steps.
nprojint > 0, optional: How often to reproject the sampling point into the feasibility space. Avoids numerical issues at the cost of lower sampling. If you observe many equality constraint violations with sampler.validate you should lower this number (default None).
seedint > 0, optional: Sets the random number seed. Initialized to the current time stamp if None (default None).

feasibility_tol: float: The tolerance used for checking equalities feasibility.
bounds_tol: float: The tolerance used for checking bounds feasibility.
n_samplesint: The total number of samples that have been generated by this sampler instance.
retriesint: The overall of sampling retries the sampler has observed. Larger values indicate numerical instabilities.
problemProblem: A NamedTuple whose attributes define the entire sampling problem in matrix form.
warmupnumpy.matrix: A numpy matrix with as many columns as reactions in the model and more than 3 rows containing a warmup sample in each row. None if no warmup points have been generated yet.
fwd_idxnumpy.array: A numpy array having one entry for each reaction in the model, containing the index of the respective forward variable.
rev_idxnumpy.array: A numpy array having one entry for each reaction in the model, containing the index of the respective reverse variable.

generate_fva_warmup(self)¶

Generate the warmup points for the sampler.

Generates warmup points by setting each flux as the sole objective and minimizing/maximizing it. Also caches the projection of the warmup points into the nullspace for non-homogeneous problems (only if necessary).

class pytfa.analysis.GeneralizedACHRSampler(model, thinning=100, seed=None)[source]¶

Bases: GeneralizedHRSampler, cobra.sampling.ACHRSampler

The abstract base class for hit-and-run samplers.

New samplers should derive from this class where possible to provide a uniform interface.

modelcobra.Model: The cobra model from which to generate samples.
thinningint: The thinning factor of the generated sampling chain. A thinning of 10 means samples are returned every 10 steps.
nprojint > 0, optional: How often to reproject the sampling point into the feasibility space. Avoids numerical issues at the cost of lower sampling. If you observe many equality constraint violations with sampler.validate you should lower this number (default None).
seedint > 0, optional: Sets the random number seed. Initialized to the current time stamp if None (default None).

feasibility_tol: float: The tolerance used for checking equalities feasibility.
bounds_tol: float: The tolerance used for checking bounds feasibility.
n_samplesint: The total number of samples that have been generated by this sampler instance.
retriesint: The overall of sampling retries the sampler has observed. Larger values indicate numerical instabilities.
problemProblem: A NamedTuple whose attributes define the entire sampling problem in matrix form.
warmupnumpy.matrix: A numpy matrix with as many columns as reactions in the model and more than 3 rows containing a warmup sample in each row. None if no warmup points have been generated yet.
fwd_idxnumpy.array: A numpy array having one entry for each reaction in the model, containing the index of the respective forward variable.
rev_idxnumpy.array: A numpy array having one entry for each reaction in the model, containing the index of the respective reverse variable.

class pytfa.analysis.GeneralizedOptGPSampler(model, processes, thinning=100, seed=None)[source]¶

Bases: GeneralizedHRSampler, cobra.sampling.OptGPSampler

The abstract base class for hit-and-run samplers.

New samplers should derive from this class where possible to provide a uniform interface.

modelcobra.Model: The cobra model from which to generate samples.
thinningint: The thinning factor of the generated sampling chain. A thinning of 10 means samples are returned every 10 steps.
nprojint > 0, optional: How often to reproject the sampling point into the feasibility space. Avoids numerical issues at the cost of lower sampling. If you observe many equality constraint violations with sampler.validate you should lower this number (default None).
seedint > 0, optional: Sets the random number seed. Initialized to the current time stamp if None (default None).

feasibility_tol: float: The tolerance used for checking equalities feasibility.
bounds_tol: float: The tolerance used for checking bounds feasibility.
n_samplesint: The total number of samples that have been generated by this sampler instance.
retriesint: The overall of sampling retries the sampler has observed. Larger values indicate numerical instabilities.
problemProblem: A NamedTuple whose attributes define the entire sampling problem in matrix form.
warmupnumpy.matrix: A numpy matrix with as many columns as reactions in the model and more than 3 rows containing a warmup sample in each row. None if no warmup points have been generated yet.
fwd_idxnumpy.array: A numpy array having one entry for each reaction in the model, containing the index of the respective forward variable.
rev_idxnumpy.array: A numpy array having one entry for each reaction in the model, containing the index of the respective reverse variable.

pytfa.analysis.sample(model, n, method='optgp', thinning=100, processes=1, seed=None)[source]¶

Sample valid flux distributions from a thermo cobra_model.

Function adapted from cobra.flux_analysis.sample to display all solver variables

Documentation adapted from cobra.flux_analysis.sample

‘optgp’ (default) which uses the OptGPSampler that supports parallel
analysis 1. Requires large numbers of samples to be performant (n < 1000). For smaller samples ‘achr’ might be better suited.

or

‘achr’ which uses artificial centering hit-and-run. This is a single process method with good convergence 2.

modelpytfa.core.ThermoModel: The cobra_model from which to sample variables.
nint: The number of samples to obtain. When using ‘optgp’ this must be a multiple of processes, otherwise a larger number of samples will be returned.
methodstr, optional: The analysis algorithm to use.
thinningint, optional: The thinning factor of the generated analysis chain. A thinning of 10 means samples are returned every 10 steps. Defaults to 100 which in benchmarks gives approximately uncorrelated samples. If set to one will return all iterates.
processesint, optional: Only used for ‘optgp’. The number of processes used to generate samples.
seedpositive integer, optional: The random number seed to be used. Initialized to current time stamp if None.

pandas.DataFrame: The generated flux samples. Each row corresponds to a sample of the fluxes and the columns are the reactions.

The samplers have a correction method to ensure equality feasibility for long-running chains, however this will only work for homogeneous models, meaning models with no non-zero fixed variables or constraints ( right-hand side of the equalities are zero).

1: Megchelenbrink W, Huynen M, Marchiori E (2014) optGpSampler: An Improved Tool for Uniformly Sampling the Solution-Space of Genome-Scale Metabolic Networks. PLoS ONE 9(2): e86587.
2: Direction Choice for Accelerated Convergence in Hit-and-Run Sampling David E. Kaufman Robert L. Smith Operations Research 199846:1 , 84-95

class pytfa.analysis.DeltaG(reaction, **kwargs)¶

Bases: ReactionVariable

Class to represent a DeltaG

prefix = DG_¶

class pytfa.analysis.ForbiddenProfile(model, expr, id_, **kwargs)¶

Bases: GenericConstraint

Class to represent a forbidden net flux directionality profile Looks like: FU_rxn_1 + BU_rxn_2 + … + FU_rxn_n <= n-1

prefix = FP_¶

pytfa.analysis.get_direction_use_variables(tmodel, solution)¶

Returns the active use variables in a solution. Use Variables are binary variables that control the directionality of the reaction The difference with get_active_use_variables is that variables with both UseVariables at 0 will return as going forwards. This is to ensure that the output size of the function is equal to the number of FDPs

ex: FU_ACALDt BU_PFK

Parameters

tmodel (pytfa.core.ThermoModel) –
solution –

Returns

pytfa.analysis.get_active_use_variables(tmodel, solution)¶

Returns the active use variables in a solution. Use Variables are binary variables that control the directionality of the reaction

ex: FU_ACALDt BU_PFK

Parameters

tmodel (pytfa.core.ThermoModel) –
solution –

Returns

class pytfa.analysis.ForwardUseVariable(reaction, **kwargs)¶

Bases: ReactionVariable, BinaryVariable

Class to represent a forward use variable, a type of binary variable used to enforce forward directionality in reaction net fluxes

prefix = FU_¶

pytfa.analysis.get_bistream_logger(name)¶

Sets up a logger that outputs INFO+ messages on stdout and DEBUG+ messages in the log file

Parameters: name – a class __name__ attribute
Returns

pytfa.analysis.CPU_COUNT[source]¶

pytfa.analysis.BEST_THREAD_RATIO[source]¶

pytfa.analysis.find_bidirectional_reactions(va, tolerance=1e-08)[source]¶

Returns the ids of reactions that can both carry net flux in the forward or backward direction.

Parameters

va –

A variability analysis, pandas Dataframe like so:: maximum minimum

6PGLter -8.330667e-04 -8.330667e-04 ABUTt2r 0.000000e+00 0.000000e+00 ACALDt 0.000000e+00 0.000000e+00

Returns

pytfa.analysis.find_directionality_profiles(tmodel, bidirectional, max_iter=10000.0, solver='optlang-glpk', tolerance=1e-09)[source]¶

Takes a ThermoModel and performs enumeration of the directionality profiles

Parameters

tmodel –
max_iter –

Returns

pytfa.analysis._bool2str(bool_list)[source]¶

turns a list of booleans into a string

Parameters: bool_list – ex: ‘[False True False False True]’
Returns: ‘01001’

pytfa.analysis._variability_analysis_element(tmodel, var, sense)[source]¶

pytfa.analysis.variability_analysis(tmodel, kind='reactions', proc_num=BEST_THREAD_RATIO)[source]¶

Performs variability analysis, gicven a variable type

Parameters

tmodel –
kind –
proc_num –

Returns

pytfa.analysis.parallel_variability_analysis(tmodel, kind='reactions', proc_num=BEST_THREAD_RATIO)[source]¶

WIP.

Parameters

tmodel –
kind –
proc_num –

Returns

pytfa.analysis.calculate_dissipation(tmodel, solution=None)[source]¶

pytfa.analysis.apply_reaction_variability(tmodel, va, inplace=True)[source]¶: Applies the VA results as bounds for the reactions of a cobra_model :param inplace: :param tmodel: :param va: :return:

pytfa.analysis.apply_generic_variability(tmodel, va, inplace=True)[source]¶

Reactions a dealt with cobra, but the other variables added use pytfa’s interface: the class GenericVariable. We use a different method to apply variability directly in the solver

Parameters

tmodel –
va –
inplace –

Returns

pytfa.analysis.apply_directionality(tmodel, solution, inplace=True)[source]¶

Takes a flux solution and transfers its reaction directionality as constraints for the cobra_model

Parameters

inplace –
tmodel –
solution –

Returns

`pytfa.core`¶

Submodules¶

pytfa.core.model¶

Model class

Module Contents¶

Classes¶

LCSBModel

Helper class that provides a standard way to create an ABC using

Functions¶

timeit(method)

Adapted from Andreas Jung's `blog

pytfa.timeit(method)¶

Adapted from Andreas Jung’s blog

Parameters: method – The method to time
Returns

class pytfa.LCSBModel(model, name, sloppy=False)¶

Bases: abc.ABC

Helper class that provides a standard way to create an ABC using inheritance.

abstract copy(self)¶: Needs to be reimplemented, as our objects have complicated hierarchy :return:

print_info(self)¶: Print information and counts for the cobra_model :return:

add_variable(self, kind, hook, queue=False, **kwargs)¶

Add a new variable to a COBRApy cobra_model.

Parameters

kind –
hook (string,cobra.Reaction) – Either a string representing the name of the variable to add to the cobra_model, or a reaction object if the kind allows it

Returns

The created variable

Return type

optlang.interface.Variable

add_constraint(self, kind, hook, expr, queue=False, **kwargs)¶

Add a new constraint to a COBRApy cobra_model

Parameters

kind –
hook (string,cobra.Reaction) – Either a string representing the name of the variable to add to the cobra_model, or a reaction object if the kind allows it
expr (sympy.thermo.expr.Expr) – The expression of the constraint

Returns

The created constraint

Return type

optlang.interface.Constraint

remove_reactions(self, reactions, remove_orphans=False)¶

remove_metabolites(self, metabolite_list, destructive=False)¶

_remove_associated_consvar(self, all_cons_subclasses, all_var_subclasses, collection)¶: Removes both the constraints and variables associated to an element, as long as it was used as a hook in the cons/var declaration. For example, upon removing a reaction, also removes its associated deltaG variables and coupling constraints

remove_variable(self, var)¶

Removes a variable

Parameters: var –
Returns

remove_constraint(self, cons)¶

Removes a constraint

Parameters: cons –
Returns

_push_queue(self)¶: updates the constraints and variables of the model with what’s in the queue :return:

regenerate_variables(self)¶: Generates references to the cobra_model’s constraints in self._var_dict as tab-searchable attributes of the thermo cobra_model :return:

regenerate_constraints(self)¶: Generates references to the cobra_model’s constraints in self._cons_dict as tab-searchable attributes of the thermo cobra_model :return:

repair(self)¶: Updates references to variables and constraints :return:

get_primal(self, vartype, index_by_reactions=False)¶

Returns the primal value of the cobra_model for variables of a given type

Parameters

index_by_reactions –
vartype – Class of variable. Ex: pytfa.optim.variables.ThermoDisplacement

Returns

get_solution(self)¶

Overrides the cobra.thermo.solution method, to also get the supplementary variables we added to the cobra_model

solution.fluxes in cobrapy is a transformed version of the solver output, as it actually calculates the _net_ flux of each reaction by substracting the reverse variable value to the forward variable value. This should be used anytime one needs the actual flux value
solution.raw is a clear copy of the solver output. From there one can access the value at solution for all the variables of the problem. However, looking for a reaction ID in there will only give the _forward_ flux. This should be used for any other variable than fluxes.
solution.values yields variables multiplied by their scaling factor (1 by default). Useful if you operated scaling on your equations for numerical reasons. This does _not_ include fluxes

Returns

optimize(self, objective_sense=None, **kwargs)¶: Call the Model.optimize function (which really is but an interface to the solver’s. Catches SolverError in the case of no solutions. Passes down supplementary keyword arguments (see cobra.thermo.Model.optimize) :type objective_sense: ‘min’ or ‘max’

slim_optimize(self, *args, **kwargs)¶

get_constraints_of_type(self, constraint_type)¶

Convenience function that takes as input a constraint class and returns all its instances within the cobra_model

Parameters: constraint_type –
Returns

get_variables_of_type(self, variable_type)¶

Convenience function that takes as input a variable class and returns all its instances within the cobra_model

Parameters: variable_type –
Returns

Package Contents¶

Classes¶

LCSBModel

Helper class that provides a standard way to create an ABC using

class pytfa.core.LCSBModel(model, name, sloppy=False)[source]¶

Bases: abc.ABC

Helper class that provides a standard way to create an ABC using inheritance.

abstract copy(self)¶: Needs to be reimplemented, as our objects have complicated hierarchy :return:

print_info(self)¶: Print information and counts for the cobra_model :return:

add_variable(self, kind, hook, queue=False, **kwargs)¶

Add a new variable to a COBRApy cobra_model.

Parameters

kind –
hook (string,cobra.Reaction) – Either a string representing the name of the variable to add to the cobra_model, or a reaction object if the kind allows it

Returns

The created variable

Return type

optlang.interface.Variable

add_constraint(self, kind, hook, expr, queue=False, **kwargs)¶

Add a new constraint to a COBRApy cobra_model

Parameters

kind –
hook (string,cobra.Reaction) – Either a string representing the name of the variable to add to the cobra_model, or a reaction object if the kind allows it
expr (sympy.thermo.expr.Expr) – The expression of the constraint

Returns

The created constraint

Return type

optlang.interface.Constraint

remove_reactions(self, reactions, remove_orphans=False)¶

remove_metabolites(self, metabolite_list, destructive=False)¶

_remove_associated_consvar(self, all_cons_subclasses, all_var_subclasses, collection)¶: Removes both the constraints and variables associated to an element, as long as it was used as a hook in the cons/var declaration. For example, upon removing a reaction, also removes its associated deltaG variables and coupling constraints

remove_variable(self, var)¶

Removes a variable

Parameters: var –
Returns

remove_constraint(self, cons)¶

Removes a constraint

Parameters: cons –
Returns

_push_queue(self)¶: updates the constraints and variables of the model with what’s in the queue :return:

regenerate_variables(self)¶: Generates references to the cobra_model’s constraints in self._var_dict as tab-searchable attributes of the thermo cobra_model :return:

regenerate_constraints(self)¶: Generates references to the cobra_model’s constraints in self._cons_dict as tab-searchable attributes of the thermo cobra_model :return:

repair(self)¶: Updates references to variables and constraints :return:

get_primal(self, vartype, index_by_reactions=False)¶

Returns the primal value of the cobra_model for variables of a given type

Parameters

index_by_reactions –
vartype – Class of variable. Ex: pytfa.optim.variables.ThermoDisplacement

Returns

get_solution(self)¶

Overrides the cobra.thermo.solution method, to also get the supplementary variables we added to the cobra_model

solution.fluxes in cobrapy is a transformed version of the solver output, as it actually calculates the _net_ flux of each reaction by substracting the reverse variable value to the forward variable value. This should be used anytime one needs the actual flux value
solution.raw is a clear copy of the solver output. From there one can access the value at solution for all the variables of the problem. However, looking for a reaction ID in there will only give the _forward_ flux. This should be used for any other variable than fluxes.
solution.values yields variables multiplied by their scaling factor (1 by default). Useful if you operated scaling on your equations for numerical reasons. This does _not_ include fluxes

Returns

optimize(self, objective_sense=None, **kwargs)¶: Call the Model.optimize function (which really is but an interface to the solver’s. Catches SolverError in the case of no solutions. Passes down supplementary keyword arguments (see cobra.thermo.Model.optimize) :type objective_sense: ‘min’ or ‘max’

slim_optimize(self, *args, **kwargs)¶

get_constraints_of_type(self, constraint_type)¶

Convenience function that takes as input a constraint class and returns all its instances within the cobra_model

Parameters: constraint_type –
Returns

get_variables_of_type(self, variable_type)¶

Convenience function that takes as input a variable class and returns all its instances within the cobra_model

Parameters: variable_type –
Returns

`pytfa.io`¶

Submodules¶

pytfa.io.base¶

Input/Output tools to import or export pytfa models

Module Contents¶

Functions¶

`import_matlab_model`(path, variable_name=None)	Convert at matlab cobra_model to a pyTFA cobra_model, with Thermodynamic values
`recover_compartments`(model, compartments_list)
`write_matlab_model`(tmodel, path, varname='tmodel')	Writes the Thermo Model to a Matlab-compatible structure
`create_thermo_dict`(tmodel)	Dumps the thermodynamic information in a mat-compatible dictionary
`varnames2matlab`(name, tmodel)	Transforms reaction variable pairs from ('ACALD','ACALD_reverse_xxxxx') to
`create_problem_dict`(tmodel)	Dumps the the MILP formulation for TFA in a mat-compatible dictionary
`create_generalized_matrix`(tmodel, array_type='dense')	Returns the generalized stoichiomatric matrix used for TFA
`load_thermoDB`(path)	Load a thermodynamic database
`printLP`(model)	Print the LP file corresponding to the cobra_model
`writeLP`(model, path=None)	Write the LP file of the specified cobra_model to the file indicated by path.

pytfa.import_matlab_model(path, variable_name=None)¶

Convert at matlab cobra_model to a pyTFA cobra_model, with Thermodynamic values

Parameters

variable_name –
path (string) – The path of the file to import

Returns

The converted cobra_model

Return type

cobra.thermo.model.Model

pytfa.recover_compartments(model, compartments_list)¶

pytfa.write_matlab_model(tmodel, path, varname='tmodel')¶

Writes the Thermo Model to a Matlab-compatible structure

Parameters

varname –
tmodel –
path –

Returns

None

pytfa.create_thermo_dict(tmodel)¶

Dumps the thermodynamic information in a mat-compatible dictionary (similar to the output of cobra.io.mat.create_mat_dict)

Parameters: tmodel – pytfa.thermo.tmodel.ThermoModel
Returns: dict object

pytfa.varnames2matlab(name, tmodel)¶

Transforms reaction variable pairs from (‘ACALD’,’ACALD_reverse_xxxxx’) to (‘F_ACALD’,’B_ACALD’) if it is a reaction, else leaves is as is

Returns

pytfa.create_problem_dict(tmodel)¶

Dumps the the MILP formulation for TFA in a mat-compatible dictionary (similar to the output of cobra.io.mat.create_mat_dict)

Parameters: tmodel – pytfa.thermo.tmodel.ThermoModel

:ret

pytfa.create_generalized_matrix(tmodel, array_type='dense')¶

Returns the generalized stoichiomatric matrix used for TFA

Parameters

array_type –
tmodel – pytfa.ThermoModel

Returns

matrix.

pytfa.load_thermoDB(path)¶

Load a thermodynamic database

Parameters: path (string) – The path of the file to load
Returns: The thermodynamic database
Return type: dict

pytfa.printLP(model)¶

Print the LP file corresponding to the cobra_model

Parameters: model (cobra.thermo.model.Model) – The cobra_model to output the LP file for
Returns: The content of the LP file
Return type: str

Usually, you pass the result of this function to file.write() to write it on disk. If you prefer, you can use pytfa.io.writeLP() to write the result directly to a file.

pytfa.writeLP(model, path=None)¶

Write the LP file of the specified cobra_model to the file indicated by path.

Parameters

model (cobra.thermo.model.Model) – The COBRApy cobra_model to write the LP file for
path (string) – Optional The path of the file to be written. If not specified, the name of the COBRApy cobra_model will be used.

pytfa.io.enrichment¶

Tools to import or export enrichment to and from pytfa models

Module Contents¶

Functions¶

`write_lexicon`(tmodel, filepath)	Writes a csv file in the format :
`annotate_from_lexicon`(model, lexicon)	Converts a lexicon into annotation for the metabolites
`read_lexicon`(filepath)
`write_compartment_data`(tmodel, filepath)	param filepath
`read_compartment_data`(filepath)
`apply_compartment_data`(tmodel, compartment_data)

pytfa.write_lexicon(tmodel, filepath)¶

Writes a csv file in the format :: seed_id

13BDglcn_c cpd11791 13dpg_c cpd00203 2pg_c cpd00482 3pg_c cpd00169 4abut_c cpd00281

Useful for exporting an annotation

Parameters

tmodel (pytfa.core.ThermoModel) –
filepath –

Returns

pytfa.annotate_from_lexicon(model, lexicon)¶

Converts a lexicon into annotation for the metabolites

Parameters

model (cobra.Model) –
lexicon –

Returns

pytfa.read_lexicon(filepath)¶

pytfa.write_compartment_data(tmodel, filepath)¶

Parameters

filepath –
tmodel (pytfa.core.ThermoModel) –

Returns

pytfa.read_compartment_data(filepath)¶

pytfa.apply_compartment_data(tmodel, compartment_data)¶

pytfa.io.json¶

JSON serialization

Module Contents¶

Classes¶

MyEncoder

We define an encoder that takes care of the serialization of numpy types,

Functions¶

`check_json_extension`(filepath)
`save_json_model`(model, filepath)
`load_json_model`(filepath)
`json_dumps_model`(model)	Returns a JSON dump as a string
`json_loads_model`(s)	Loads a model from a string JSON dump

class pytfa.MyEncoder(*, skipkeys=False, ensure_ascii=True, check_circular=True, allow_nan=True, sort_keys=False, indent=None, separators=None, default=None)¶

Bases: json.JSONEncoder

We define an encoder that takes care of the serialization of numpy types, which are not handled by json by default

default(self, obj)¶

Implement this method in a subclass such that it returns a serializable object for o, or calls the base implementation (to raise a TypeError).

For example, to support arbitrary iterators, you could implement default like this:

def default(self, o):
    try:
        iterable = iter(o)
    except TypeError:
        pass
    else:
        return list(iterable)
    # Let the base class default method raise the TypeError
    return JSONEncoder.default(self, o)

pytfa.check_json_extension(filepath)¶

pytfa.save_json_model(model, filepath)¶

pytfa.load_json_model(filepath)¶

pytfa.json_dumps_model(model)¶

Returns a JSON dump as a string

Parameters: model –
Returns

pytfa.json_loads_model(s)¶

Loads a model from a string JSON dump

Parameters: s – JSON string
Returns

pytfa.io.plotting¶

Plotting results

Module Contents¶

Functions¶

`plot_fva_tva_comparison`(fva, tva)
`plot_thermo_displacement_histogram`(tmodel, solution=None)	Plot a histogram of the thermodynamic displacement. if no solution is
`plot_histogram`(values, **kwargs)	Convenience function. Plots a histogram of flat 1D data.

pytfa.plot_fva_tva_comparison(fva, tva)¶

pytfa.plot_thermo_displacement_histogram(tmodel, solution=None)¶

Plot a histogram of the thermodynamic displacement. if no solution is provided, will look at the cobra_model’s own solution

Parameters

tmodel –
solution –

Returns

pytfa.plot_histogram(values, **kwargs)¶

Convenience function. Plots a histogram of flat 1D data.

Parameters: values –
Returns

pytfa.io.viz¶

Input/Output tools to vizualize results

Module Contents¶

Functions¶

`export_variable_for_escher`(tmodel, variable_type, data, filename)	Exports all the variables of a given type into a csv file, indexed by
`get_reaction_data`(tmodel, data)	Exports values indexed by reaction ids. Reconciles Forward and Backwards
`export_reactions_for_escher`(tmodel, data, filename)	Exports values indexed by reaction ids. Reconciles Forward and Backwards

pytfa.export_variable_for_escher(tmodel, variable_type, data, filename)¶

Exports all the variables of a given type into a csv file, indexed by variable.id. This format is readable by escher if the variable_type is a subclass of :pytfa:`pytfa.optim.variables.ReactionVariable` or :pytfa:`pytfa.optim.variables.MetaboliteVariable`

Parameters

tmodel (pytfa.core.ThermoModel) –
variable_type (ReactionVariable|MetaboliteVariable) –
data (pandas.Series) – indexed by variable name
filename (string) –

Returns

pytfa.get_reaction_data(tmodel, data)¶: Exports values indexed by reaction ids. Reconciles Forward and Backwards variables.

pytfa.export_reactions_for_escher(tmodel, data, filename)¶

Exports values indexed by reaction ids. Reconciles Forward and Backwards variables. Writes it in a csv file. This format is readable by escher

Parameters

tmodel (pytfa.core.ThermoModel) –
variable_type (ReactionVariable|MetaboliteVariable) –
data (pandas.Series) – indexed by variable name
filename (string) –

Returns

Package Contents¶

Classes¶

MyEncoder

We define an encoder that takes care of the serialization of numpy types,

Functions¶

`import_matlab_model`(path, variable_name=None)	Convert at matlab cobra_model to a pyTFA cobra_model, with Thermodynamic values
`recover_compartments`(model, compartments_list)
`write_matlab_model`(tmodel, path, varname='tmodel')	Writes the Thermo Model to a Matlab-compatible structure
`create_thermo_dict`(tmodel)	Dumps the thermodynamic information in a mat-compatible dictionary
`varnames2matlab`(name, tmodel)	Transforms reaction variable pairs from ('ACALD','ACALD_reverse_xxxxx') to
`create_problem_dict`(tmodel)	Dumps the the MILP formulation for TFA in a mat-compatible dictionary
`create_generalized_matrix`(tmodel, array_type='dense')	Returns the generalized stoichiomatric matrix used for TFA
`load_thermoDB`(path)	Load a thermodynamic database
`printLP`(model)	Print the LP file corresponding to the cobra_model
`writeLP`(model, path=None)	Write the LP file of the specified cobra_model to the file indicated by path.
`write_lexicon`(tmodel, filepath)	Writes a csv file in the format :
`annotate_from_lexicon`(model, lexicon)	Converts a lexicon into annotation for the metabolites
`read_lexicon`(filepath)
`write_compartment_data`(tmodel, filepath)	param filepath
`read_compartment_data`(filepath)
`apply_compartment_data`(tmodel, compartment_data)

Attributes¶

BIGM_DG

pytfa.io.BIGM_DG = 1000.0¶

pytfa.io.import_matlab_model(path, variable_name=None)[source]¶

Convert at matlab cobra_model to a pyTFA cobra_model, with Thermodynamic values

Parameters

variable_name –
path (string) – The path of the file to import

Returns

The converted cobra_model

Return type

cobra.thermo.model.Model

pytfa.io.recover_compartments(model, compartments_list)[source]¶

pytfa.io.write_matlab_model(tmodel, path, varname='tmodel')[source]¶

Writes the Thermo Model to a Matlab-compatible structure

Parameters

varname –
tmodel –
path –

Returns

None

pytfa.io.create_thermo_dict(tmodel)[source]¶

Dumps the thermodynamic information in a mat-compatible dictionary (similar to the output of cobra.io.mat.create_mat_dict)

Parameters: tmodel – pytfa.thermo.tmodel.ThermoModel
Returns: dict object

pytfa.io.varnames2matlab(name, tmodel)[source]¶

Transforms reaction variable pairs from (‘ACALD’,’ACALD_reverse_xxxxx’) to (‘F_ACALD’,’B_ACALD’) if it is a reaction, else leaves is as is

Returns

pytfa.io.create_problem_dict(tmodel)[source]¶

Dumps the the MILP formulation for TFA in a mat-compatible dictionary (similar to the output of cobra.io.mat.create_mat_dict)

Parameters: tmodel – pytfa.thermo.tmodel.ThermoModel

:ret

pytfa.io.create_generalized_matrix(tmodel, array_type='dense')[source]¶

Returns the generalized stoichiomatric matrix used for TFA

Parameters

array_type –
tmodel – pytfa.ThermoModel

Returns

matrix.

pytfa.io.load_thermoDB(path)[source]¶

Load a thermodynamic database

Parameters: path (string) – The path of the file to load
Returns: The thermodynamic database
Return type: dict

pytfa.io.printLP(model)[source]¶

Print the LP file corresponding to the cobra_model

Parameters: model (cobra.thermo.model.Model) – The cobra_model to output the LP file for
Returns: The content of the LP file
Return type: str

Usually, you pass the result of this function to file.write() to write it on disk. If you prefer, you can use pytfa.io.writeLP() to write the result directly to a file.

pytfa.io.writeLP(model, path=None)[source]¶

Write the LP file of the specified cobra_model to the file indicated by path.

Parameters

model (cobra.thermo.model.Model) – The COBRApy cobra_model to write the LP file for
path (string) – Optional The path of the file to be written. If not specified, the name of the COBRApy cobra_model will be used.

class pytfa.io.MyEncoder(*, skipkeys=False, ensure_ascii=True, check_circular=True, allow_nan=True, sort_keys=False, indent=None, separators=None, default=None)¶

Bases: json.JSONEncoder

We define an encoder that takes care of the serialization of numpy types, which are not handled by json by default

default(self, obj)¶

Implement this method in a subclass such that it returns a serializable object for o, or calls the base implementation (to raise a TypeError).

For example, to support arbitrary iterators, you could implement default like this:

def default(self, o):
    try:
        iterable = iter(o)
    except TypeError:
        pass
    else:
        return list(iterable)
    # Let the base class default method raise the TypeError
    return JSONEncoder.default(self, o)

pytfa.io.write_lexicon(tmodel, filepath)[source]¶

Writes a csv file in the format :: seed_id

13BDglcn_c cpd11791 13dpg_c cpd00203 2pg_c cpd00482 3pg_c cpd00169 4abut_c cpd00281

Useful for exporting an annotation

Parameters

tmodel (pytfa.core.ThermoModel) –
filepath –

Returns

pytfa.io.annotate_from_lexicon(model, lexicon)[source]¶

Converts a lexicon into annotation for the metabolites

Parameters

model (cobra.Model) –
lexicon –

Returns

pytfa.io.read_lexicon(filepath)[source]¶

pytfa.io.write_compartment_data(tmodel, filepath)[source]¶

Parameters

filepath –
tmodel (pytfa.core.ThermoModel) –

Returns

pytfa.io.read_compartment_data(filepath)[source]¶

pytfa.io.apply_compartment_data(tmodel, compartment_data)[source]¶

`pytfa.optim`¶

Submodules¶

pytfa.optim.config¶

Pre-tuned configurations for faster solving

Module Contents¶

Functions¶

dg_relax_config(model)

param model

pytfa.dg_relax_config(model)¶

Parameters: model –
Returns

pytfa.optim.constraints¶

Constraints declarations

Module Contents¶

Classes¶

`GenericConstraint`	Class to represent a generic constraint. The purpose is that the interface
`ModelConstraint`	Class to represent a variable attached to the model
`GeneConstraint`	Class to represent a variable attached to a enzyme
`ReactionConstraint`	Class to represent a variable attached to a reaction
`MetaboliteConstraint`	Class to represent a variable attached to a enzyme
`NegativeDeltaG`	Class to represent thermodynamics constraints.
`NegativeDeltaGIneq`	Class to represent thermodynamics constraints.
`ForwardDeltaGCoupling`	Class to represent thermodynamics coupling: DeltaG of reactions has to be
`BackwardDeltaGCoupling`	Class to represent thermodynamics coupling: DeltaG of reactions has to be
`ForwardDirectionCoupling`	Class to represent a forward directionality coupling with thermodynamics on
`BackwardDirectionCoupling`	Class to represent a backward directionality coupling with thermodynamics on
`SimultaneousUse`	Class to represent a simultaneous use constraint on reaction variables
`DisplacementCoupling`	Class to represent the coupling to the thermodynamic displacement
`ForbiddenProfile`	Class to represent a forbidden net flux directionality profile
`LinearizationConstraint`	Class to represent a variable attached to a reaction
`LowerBoundLogConcentration`	Class to represent a forbidden net flux directionality profile
`UpperBoundLogConcentration`	Class to represent a forbidden net flux directionality profile
`NullspaceConstraint`	Class to represent a forbidden net flux directionality profile
`PotentialConstraint`	Class to represent a forbidden net flux directionality profile
`PotentialCoupling`	Class to rep couple DG to potential variable

class pytfa.GenericConstraint(expr, id_='', model=None, hook=None, queue=False, **kwargs)¶

Class to represent a generic constraint. The purpose is that the interface: is instantiated on initialization, to follow the type of interface used by the problem, and avoid incompatibilities in optlang

Attributes:

id

Used for DictList comprehension. Usually points back at a

enzyme or reaction id for ease of linking. Should be unique given a constraint type. :name: Should be a concatenation of the id and a prefix that is specific to the variable type. will be used to address the constraint at the solver level, and hence should be unique in the whole cobra_model :expr: the expression of the constraint (sympy.Expression subtype) :cobra_model: the cobra_model hook. :constraint: links directly to the cobra_model representation of tbe constraint

prefix¶

property __attrname__(self)¶: Name the attribute the instances will have Example: GenericConstraint -> generic_constraint :return:

get_interface(self, expr, queue)¶

Called upon completion of __init__, initializes the value of self.var, which is returned upon call, and stores the actual interfaced variable.

Returns: instance of Variable from the problem

make_name(self)¶

Needs to be overridden by the subclass, concats the id with a: prefix

Returns: None

change_expr(self, new_expr, sloppy=False)¶

property expr(self)¶

property name(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property constraint(self)¶

property model(self)¶

__repr__(self)¶: Return repr(self).

class pytfa.ModelConstraint(model, expr, id_, **kwargs)¶

Bases: GenericConstraint

Class to represent a variable attached to the model

prefix = MODC_¶

class pytfa.GeneConstraint(gene, expr, **kwargs)¶

Bases: GenericConstraint

Class to represent a variable attached to a enzyme

prefix = GC_¶

property gene(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.ReactionConstraint(reaction, expr, **kwargs)¶

Bases: GenericConstraint

Class to represent a variable attached to a reaction

prefix = RC_¶

property reaction(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.MetaboliteConstraint(metabolite, expr, **kwargs)¶

Bases: GenericConstraint

Class to represent a variable attached to a enzyme

prefix = MC_¶

property metabolite(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.NegativeDeltaG(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent thermodynamics constraints.

G: - DGR_rxn + DGoRerr_Rxn + RT * StoichCoefProd1 * LC_prod1

RT * StoichCoefProd2 * LC_prod2

RT * StoichCoefSub1 * LC_subs1

RT * StoichCoefSub2 * LC_subs2

…

= 0

prefix = G_¶

class pytfa.NegativeDeltaGIneq(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent thermodynamics constraints.

G: DGoRerr_Rxn + RT * StoichCoefProd1 * LC_prod1

RT * StoichCoefProd2 * LC_prod2

RT * StoichCoefSub1 * LC_subs1

RT * StoichCoefSub2 * LC_subs2

…

< or > 0

prefix = GI_¶

class pytfa.ForwardDeltaGCoupling(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent thermodynamics coupling: DeltaG of reactions has to be DGR < 0 for the reaction to proceed forwards Looks like: FU_rxn: 1000 FU_rxn + DGR_rxn < 1000

prefix = FU_¶

class pytfa.BackwardDeltaGCoupling(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent thermodynamics coupling: DeltaG of reactions has to be DGR > 0 for the reaction to proceed backwards Looks like: BU_rxn: 1000 BU_rxn - DGR_rxn < 1000

prefix = BU_¶

class pytfa.ForwardDirectionCoupling(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent a forward directionality coupling with thermodynamics on reaction variables Looks like : UF_rxn: F_rxn - M FU_rxn < 0

prefix = UF_¶

class pytfa.BackwardDirectionCoupling(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent a backward directionality coupling with thermodynamics on reaction variables Looks like : UR_rxn: R_rxn - M RU_rxn < 0

prefix = UR_¶

class pytfa.SimultaneousUse(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent a simultaneous use constraint on reaction variables Looks like: SU_rxn: FU_rxn + BU_rxn <= 1

prefix = SU_¶

class pytfa.DisplacementCoupling(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent the coupling to the thermodynamic displacement Looks like: Ln(Gamma) - (1/RT)*DGR_rxn = 0

prefix = DC_¶

class pytfa.ForbiddenProfile(model, expr, id_, **kwargs)¶

Bases: GenericConstraint

Class to represent a forbidden net flux directionality profile Looks like: FU_rxn_1 + BU_rxn_2 + … + FU_rxn_n <= n-1

prefix = FP_¶

class pytfa.LinearizationConstraint(model, expr, id_, **kwargs)¶

Bases: ModelConstraint

Class to represent a variable attached to a reaction

prefix = LC_¶

static from_constraints(cons, model)¶

class pytfa.LowerBoundLogConcentration(metabolite, expr, **kwargs)¶

Bases: MetaboliteConstraint

Class to represent a forbidden net flux directionality profile Looks like: DG_rxn_1 + DG_rxn_2 + … + DG_rxn_n == 0

prefix = LB_¶

class pytfa.UpperBoundLogConcentration(metabolite, expr, **kwargs)¶

Bases: MetaboliteConstraint

Class to represent a forbidden net flux directionality profile Looks like: DG_rxn_1 + DG_rxn_2 + … + DG_rxn_n == 0

prefix = UB_¶

class pytfa.NullspaceConstraint(metabolite, expr, **kwargs)¶

Bases: MetaboliteConstraint

Class to represent a forbidden net flux directionality profile Looks like: DG_rxn_1 + DG_rxn_2 + … + DG_rxn_n == 0

prefix = NS_¶

class pytfa.PotentialConstraint(metabolite, expr, **kwargs)¶

Bases: MetaboliteConstraint

Class to represent a forbidden net flux directionality profile Looks like: DG_rxn_1 + DG_rxn_2 + … + DG_rxn_n == 0

prefix = PC_¶

class pytfa.PotentialCoupling(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to rep couple DG to potential variable

prefix = PPC_¶

pytfa.optim.debugging¶

Debugging of models

Module Contents¶

Functions¶

`debug_iis`(model)	Performs reduction to an Irreducible Inconsistent Subsystem (IIS)
`find_extreme_coeffs`(model, n=5)
`find_maxed_vars`(model, ub=1000, epsilon=0.01)

pytfa.debug_iis(model)¶

Performs reduction to an Irreducible Inconsistent Subsystem (IIS)

Parameters: model –
Returns

pytfa.find_extreme_coeffs(model, n=5)¶

pytfa.find_maxed_vars(model, ub=1000, epsilon=0.01)¶

pytfa.optim.meta¶

Metaclass declarations to force the definition of prefixes in GenericVariable and GeneriConstraint subclasses

Based on SethMMorton’s answer on StackOverflow https://stackoverflow.com/questions/45248243/most-pythonic-way-to-declare-an-abstract-class-property https://stackoverflow.com/a/45250114

Module Contents¶

Classes¶

RequirePrefixMeta

Metaclass that enforces child classes define prefix.

Attributes¶

ABCRequirePrefixMeta

class pytfa.RequirePrefixMeta(cls, name, bases, attrs)¶

Bases: type

Metaclass that enforces child classes define prefix.

pytfa.ABCRequirePrefixMeta¶

pytfa.optim.reformulation¶

MILP-fu to reformulate problems

Module Contents¶

Functions¶

`subs_bilinear`(expr)	Substitutes bilinear forms from an expression with dedicated variables
`glovers_linearization`(b, fy, z=None, L=0, U=1000)	Glover, Fred.
`petersen_linearization`(b, x, z=None, M=1000)	PETERSEN, C,,
`linearize_product`(model, b, x, queue=False)	param model

Attributes¶

`ConstraintTuple`
`OPTLANG_BINARY`

pytfa.ConstraintTuple¶

pytfa.OPTLANG_BINARY = binary¶

pytfa.subs_bilinear(expr)¶: Substitutes bilinear forms from an expression with dedicated variables :param expr: :return:

pytfa.glovers_linearization(b, fy, z=None, L=0, U=1000)¶

Glover, Fred. “Improved linear integer programming formulations of nonlinear integer problems.” Management Science 22.4 (1975): 455-460.

Performs Glovers Linearization of a product z = b*f(y) <=> z - b*f(y) = 0 <=> { L*b <= z <= U*b { f(y) - U*(1-b) <= z <= f(y) - L*(1-b)

where : * b is a binary variable * f a linear combination of continuous or integer variables y

Parameters

b – Must be a binary optlang variable
z – Must be an optlang variable. Will be mapped to the product so that z = b*f(y)
fy – Must be an expression or variable
L – minimal value for fy
U – maximal value for fy

Returns

pytfa.petersen_linearization(b, x, z=None, M=1000)¶

PETERSEN, C,, “A Note on Transforming the Product of Variables to Linear Form in Linear CLIFFORD Programs,” Working Paper, Purdue University, 1971.

Performs Petersen Linearization of a product z = b*x <=> z - b*x = 0 <=> { x + M*b - M <= z <= M*b { z <= x

where : * b is a binary variable * f a linear combination of continuous or integer variables y

Parameters

x – Must be an expression or variable
b – Must be a binary optlang variable
z – Must be an optlang variable. Will be mapped to the product so that z = b*f(y)
M – big-M constraint

Returns

pytfa.linearize_product(model, b, x, queue=False)¶

Parameters

model –
b – the binary variable
x – the continuous variable
queue – whether to queue the variables and constraints made

Returns

pytfa.optim.relaxation¶

Relaxation of models with constraint too tight

Module Contents¶

Functions¶

`relax_dgo_gurobi`(model, relax_obj_type=0)
`relax_dgo`(tmodel, reactions_to_ignore=(), solver=None, in_place=False)	param t_tmodel
`relax_dgo_err`(tmodel, reactions_to_ignore=(), max_sigma=3, solver=None, in_place=False)	param t_tmodel
`relax_lc`(tmodel, metabolites_to_ignore=(), solver=None)	param metabolites_to_ignore

Attributes¶

`BIGM`
`BIGM_THERMO`
`BIGM_DG`
`BIGM_P`
`EPSILON`

pytfa.BIGM¶

pytfa.BIGM_THERMO¶

pytfa.BIGM_DG¶

pytfa.BIGM_P¶

pytfa.EPSILON¶

pytfa.relax_dgo_gurobi(model, relax_obj_type=0)¶

pytfa.relax_dgo(tmodel, reactions_to_ignore=(), solver=None, in_place=False)¶

Parameters

t_tmodel (pytfa.thermo.ThermoModel:) –
reactions_to_ignore – Iterable of reactions that should not be relaxed
solver – solver to use (e.g. ‘optlang-glpk’, ‘optlang-cplex’, ‘optlang-gurobi’

Returns

a cobra_model with relaxed bounds on standard Gibbs free energy

pytfa.relax_dgo_err(tmodel, reactions_to_ignore=(), max_sigma=3, solver=None, in_place=False)¶

Parameters

t_tmodel (pytfa.thermo.ThermoModel:) –
reactions_to_ignore – Iterable of reactions that should not be relaxed
solver – solver to use (e.g. ‘optlang-glpk’, ‘optlang-cplex’, ‘optlang-gurobi’

Returns

a cobra_model with relaxed bounds on standard Gibbs free energy

pytfa.relax_lc(tmodel, metabolites_to_ignore=(), solver=None)¶

Parameters

metabolites_to_ignore –
in_tmodel (pytfa.thermo.ThermoModel:) –
min_objective_value –

Returns

pytfa.optim.utils¶

Relaxation of models with constraint too tight

Module Contents¶

Functions¶

`get_all_subclasses`(cls)	Given a variable or constraint class, get all the subclassses
`chunk_sum`(variables)	This functions handles the sum of many sympy variables by chunks, which
`symbol_sum`(variables)	``` python
`get_solution_value_for_variables`(solution, these_vars, index_by_reaction=False)
`compare_solutions`(models)	returns the solution dictionnary for each cobra_model
`evaluate_constraint_at_solution`(constraint, solution)	param expression
`get_active_use_variables`(tmodel, solution)	Returns the active use variables in a solution. Use Variables are binary
`get_direction_use_variables`(tmodel, solution)	Returns the active use variables in a solution. Use Variables are binary
`get_primal`(tmodel, vartype, index_by_reactions=False)	Returns the primal value of the cobra_model for variables of a given type
`strip_from_integer_variables`(tmodel)	Removes all integer and binary variables of a cobra_model, to make it sample-able
`copy_solver_configuration`(source, target)	Copies the solver configuration from a source model to a target model

Attributes¶

`SYMPY_ADD_CHUNKSIZE`
`INTEGER_VARIABLE_TYPES`

pytfa.SYMPY_ADD_CHUNKSIZE = 100¶

pytfa.INTEGER_VARIABLE_TYPES = ['binary', 'integer']¶

pytfa.get_all_subclasses(cls)¶

Given a variable or constraint class, get all the subclassses that inherit from it

Parameters: cls –
Returns

pytfa.chunk_sum(variables)¶

This functions handles the sum of many sympy variables by chunks, which somehow increases the speed of the computation

You can test it in IPython: ```python a = sympy.symbols(‘a0:100’) %timeit (sum(a)) # >>> 198 µs ± 11.4 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)

b = sympy.symbols(‘b0:1000’) %timeit (sum(b)) # >>> 1.85 ms ± 356 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)

c = sympy.symbols(‘c0:3000’) %timeit (sum(c)) # >>> 5min 7s ± 2.57 s per loop (mean ± std. dev. of 7 runs, 1 loop each) ```

See the github thread

Parameters: variables –
Returns

pytfa.symbol_sum(variables)¶

``` python a = symbols(‘a0:100’)

%timeit Add(*a) # >>> 10000 loops, best of 3: 34.1 µs per loop

b = symbols(‘b0:1000’)

%timeit Add(*b) # >>> 1000 loops, best of 3: 343 µs per loop

c = symbols(‘c0:3000’)

%timeit Add(*c) # >>> 1 loops, best of 3: 1.03 ms per loop ```

See the github thread :param variables: :return:

pytfa.get_solution_value_for_variables(solution, these_vars, index_by_reaction=False)¶

pytfa.compare_solutions(models)¶: returns the solution dictionnary for each cobra_model :param (iterable (pytfa.thermo.ThermoModel)) models: :return:

pytfa.evaluate_constraint_at_solution(constraint, solution)¶

Parameters

expression –
solution – pandas.DataFrame, with index as variable names

Returns

pytfa.get_active_use_variables(tmodel, solution)¶

Returns the active use variables in a solution. Use Variables are binary variables that control the directionality of the reaction

ex: FU_ACALDt BU_PFK

Parameters

tmodel (pytfa.core.ThermoModel) –
solution –

Returns

pytfa.get_direction_use_variables(tmodel, solution)¶

Returns the active use variables in a solution. Use Variables are binary variables that control the directionality of the reaction The difference with get_active_use_variables is that variables with both UseVariables at 0 will return as going forwards. This is to ensure that the output size of the function is equal to the number of FDPs

ex: FU_ACALDt BU_PFK

Parameters

tmodel (pytfa.core.ThermoModel) –
solution –

Returns

pytfa.get_primal(tmodel, vartype, index_by_reactions=False)¶

Returns the primal value of the cobra_model for variables of a given type :param tmodel: :param vartype: Class of variable. Ex: pytfa.optim.variables.ThermoDisplacement :param index_by_reactions: Set to true to get reaction names as index instead of

variables. Useful for Escher export

Returns

pytfa.strip_from_integer_variables(tmodel)¶: Removes all integer and binary variables of a cobra_model, to make it sample-able :param tmodel: :return:

pytfa.copy_solver_configuration(source, target)¶: Copies the solver configuration from a source model to a target model :param source: :param target: :return:

pytfa.optim.variables¶

Variable declarations

Module Contents¶

Classes¶

`GenericVariable`	Class to represent a generic variable. The purpose is that the interface
`ModelVariable`	Class to represent a variable attached to the model
`GeneVariable`	Class to represent a gene variable
`BinaryVariable`	Class to represent a generic binary variable
`IntegerVariable`	Class to represent a generic binary variable
`ReactionVariable`	Class to represent a variable attached to a reaction
`MetaboliteVariable`	Class to represent a variable attached to a enzyme
`ForwardUseVariable`	Class to represent a forward use variable, a type of binary variable used to
`BackwardUseVariable`	Class to represent a backward use variable, a type of binary variable used
`ForwardBackwardUseVariable`	Class to represent a type of binary variable used to tell whether the
`LogConcentration`	Class to represent a log concentration of a enzyme
`DeltaGErr`	Class to represent a DeltaGErr
`DeltaG`	Class to represent a DeltaG
`ThermoPotential`	Class to represent a DeltaG
`DeltaGFormstd`	Class to represent a DeltaGstf of formation
`DeltaGstd`	Class to represent a DeltaG^o (naught) - standard conditions
`ThermoDisplacement`	Class to represent the thermodynamic displacement of a reaction
`PosSlackVariable`	Class to represent a positive slack variable for relaxation problems
`NegSlackVariable`	Class to represent a negative slack variable for relaxation problems
`PosSlackVariableInt`	Class to represent a postive slack variable for relaxation problems using integer
`NegSlackVariableInt`	Class to represent a postive slack variable for relaxation problems using integer
`PosSlackLC`	Class to represent a variable attached to a enzyme
`NegSlackLC`	Class to represent a variable attached to a enzyme
`LinearizationVariable`	Class to represent a product A*B when performing linearization of the

Functions¶

get_binary_type()

FIX : We enforce type to be integer instead of binary, else optlang does

Attributes¶

op_replace_dict

pytfa.op_replace_dict¶

class pytfa.GenericVariable(id_='', model=None, hook=None, queue=False, scaling_factor=1, **kwargs)¶

Class to represent a generic variable. The purpose is that the interface is instantiated on initialization, to follow the type of interface used by the problem, and avoid incompatibilities in optlang

Attributes:

id

Used for DictList comprehension. Usually points back at a

enzyme or reaction id for ease of linking. Should be unique given a variable type. :name: Should be a concatenation of the id and a prefix that is specific to the variable type. will be used to address the variable at the solver level, and hence should be unique in the whole cobra_model :cobra_model: the cobra_model hook. :variable: links directly to the cobra_model representation of tbe variable

prefix¶

property __attrname__(self)¶: Name the attribute the instances will have Example: GenericVariable -> generic_variable :return:

get_interface(self, queue)¶

Called upon completion of __init__, initializes the value of self.var, which is returned upon call, and stores the actual interfaced variable.

Returns: instance of Variable from the problem

make_name(self)¶

Needs to be overridden by the subclass, concats the id with a: prefix

Returns: None

property name(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property variable(self)¶

property scaling_factor(self)¶

property unscaled(self)¶

If the scaling factor of quantity X is a, it is represented by the variable X_hat = X/a. This returns X = a.X_hat Useful for nondimensionalisation of variables and constraints

Returns: The variable divided by its scaling factor

property value(self)¶

property unscaled_value(self)¶

property model(self)¶

property type(self)¶

test_consistency(self, other)¶

Tests whether a candidate to an operation is of the right type and is from the same problem

Parameters: other – an object
Returns: None

get_operand(self, other)¶

For operations, choose if the operand is a GenericVariable, in which we return its optlang variable, or something else (presumably a numeric) and we let optlang decide what to do

Parameters: other –
Returns

__add__(self, other)¶: Adding either two variables together or a variable and a numeric results in a new variable :param other: :return: a new Generic Variable

__radd__(self, other)¶: Take priority on symmetric arithmetic operation :param other: :return:

__sub__(self, other)¶: Substracting either two variables together or a variable and a numeric results in a new variable :param other: :return: a new Generic Variable

__rsub__(self, other)¶: Take priority on symmetric arithmetic operation :param other: :return:

__mul__(self, other)¶: Multiplying either two variables together or a variable and a numeric results in a new variable :param other: :return: a new Generic Variable

__rmul__(self, other)¶: Take priority on symmetric arithmetic operation :param other: :return:

__truediv__(self, other)¶: Dividing either two variables together or a variable and a numeric results in a new variable :param other: :return: a new Generic Variable

__rtruediv__(self, other)¶: Take priority on symmetric arithmetic operation :param other: :return:

make_result(self, new_variable)¶: Returns a Sympy expression :param new_variable: :return:

__repr__(self)¶: Return repr(self).

pytfa.get_binary_type()¶: FIX : We enforce type to be integer instead of binary, else optlang does not allow to set the binary variable bounds to anything other than (0,1) You might want to set it at (0,0) to enforce directionality for example

class pytfa.ModelVariable(model, id_, **kwargs)¶

Bases: GenericVariable

Class to represent a variable attached to the model

prefix = MODV_¶

class pytfa.GeneVariable(gene, **kwargs)¶

Bases: GenericVariable

Class to represent a gene variable

prefix = GV_¶

property gene(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.BinaryVariable(id_, model, **kwargs)¶

Bases: GenericVariable

Class to represent a generic binary variable

prefix = B_¶

class pytfa.IntegerVariable(id_, model, **kwargs)¶

Bases: GenericVariable

Class to represent a generic binary variable

prefix = I_¶

class pytfa.ReactionVariable(reaction, **kwargs)¶

Bases: GenericVariable

Class to represent a variable attached to a reaction

prefix = RV_¶

property reaction(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.MetaboliteVariable(metabolite, **kwargs)¶

Bases: GenericVariable

Class to represent a variable attached to a enzyme

prefix = MV_¶

property metabolite(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.ForwardUseVariable(reaction, **kwargs)¶

Bases: ReactionVariable, BinaryVariable

Class to represent a forward use variable, a type of binary variable used to enforce forward directionality in reaction net fluxes

prefix = FU_¶

class pytfa.BackwardUseVariable(reaction, **kwargs)¶

Bases: ReactionVariable, BinaryVariable

Class to represent a backward use variable, a type of binary variable used to enforce backward directionality in reaction net fluxes

prefix = BU_¶

class pytfa.ForwardBackwardUseVariable(reaction, **kwargs)¶

Bases: ReactionVariable, BinaryVariable

Class to represent a type of binary variable used to tell whether the reaction is active or not such that:

FU + BU + BFUSE = 1

prefix = BFUSE_¶

class pytfa.LogConcentration(metabolite, **kwargs)¶

Bases: MetaboliteVariable

Class to represent a log concentration of a enzyme

prefix = LC_¶

class pytfa.DeltaGErr(reaction, **kwargs)¶

Bases: ReactionVariable

Class to represent a DeltaGErr

prefix = DGE_¶

class pytfa.DeltaG(reaction, **kwargs)¶

Bases: ReactionVariable

Class to represent a DeltaG

prefix = DG_¶

class pytfa.ThermoPotential(metabolite, **kwargs)¶

Bases: MetaboliteVariable

Class to represent a DeltaG

prefix = P_¶

class pytfa.DeltaGFormstd(metabolite, **kwargs)¶

Bases: MetaboliteVariable

Class to represent a DeltaGstf of formation

prefix = DGoF_¶

class pytfa.DeltaGstd(reaction, **kwargs)¶

Bases: ReactionVariable

Class to represent a DeltaG^o (naught) - standard conditions

prefix = DGo_¶

class pytfa.ThermoDisplacement(reaction, **kwargs)¶

Bases: ReactionVariable

Class to represent the thermodynamic displacement of a reaction Gamma = -DeltaG/RT

prefix = LnGamma_¶

class pytfa.PosSlackVariable(reaction, **kwargs)¶

Bases: ReactionVariable

Class to represent a positive slack variable for relaxation problems

prefix = PosSlack_¶

class pytfa.NegSlackVariable(reaction, **kwargs)¶

Bases: ReactionVariable

Class to represent a negative slack variable for relaxation problems

prefix = NegSlack_¶

class pytfa.PosSlackVariableInt(reaction, **kwargs)¶

Bases: ReactionVariable, IntegerVariable

Class to represent a postive slack variable for relaxation problems using integer enforce forward directionality in reaction net fluxes

prefix = PosSlackInt_¶

class pytfa.NegSlackVariableInt(reaction, **kwargs)¶

Bases: ReactionVariable, IntegerVariable

Class to represent a postive slack variable for relaxation problems using integer enforce forward directionality in reaction net fluxes

prefix = NegSlackInt_¶

class pytfa.PosSlackLC(metabolite, **kwargs)¶

Bases: MetaboliteVariable

Class to represent a variable attached to a enzyme

prefix = PosSlackLC_¶

class pytfa.NegSlackLC(metabolite, **kwargs)¶

Bases: MetaboliteVariable

Class to represent a variable attached to a enzyme

prefix = NegSlackLC_¶

class pytfa.LinearizationVariable(model, id_, **kwargs)¶

Bases: ModelVariable

Class to represent a product A*B when performing linearization of the model

prefix = LZ_¶

Package Contents¶

Classes¶

`GenericConstraint`	Class to represent a generic constraint. The purpose is that the interface
`ModelConstraint`	Class to represent a variable attached to the model
`GeneConstraint`	Class to represent a variable attached to a enzyme
`ReactionConstraint`	Class to represent a variable attached to a reaction
`MetaboliteConstraint`	Class to represent a variable attached to a enzyme
`NegativeDeltaG`	Class to represent thermodynamics constraints.
`NegativeDeltaGIneq`	Class to represent thermodynamics constraints.
`ForwardDeltaGCoupling`	Class to represent thermodynamics coupling: DeltaG of reactions has to be
`BackwardDeltaGCoupling`	Class to represent thermodynamics coupling: DeltaG of reactions has to be
`ForwardDirectionCoupling`	Class to represent a forward directionality coupling with thermodynamics on
`BackwardDirectionCoupling`	Class to represent a backward directionality coupling with thermodynamics on
`SimultaneousUse`	Class to represent a simultaneous use constraint on reaction variables
`DisplacementCoupling`	Class to represent the coupling to the thermodynamic displacement
`ForbiddenProfile`	Class to represent a forbidden net flux directionality profile
`LinearizationConstraint`	Class to represent a variable attached to a reaction
`LowerBoundLogConcentration`	Class to represent a forbidden net flux directionality profile
`UpperBoundLogConcentration`	Class to represent a forbidden net flux directionality profile
`NullspaceConstraint`	Class to represent a forbidden net flux directionality profile
`PotentialConstraint`	Class to represent a forbidden net flux directionality profile
`PotentialCoupling`	Class to rep couple DG to potential variable
`GenericVariable`	Class to represent a generic variable. The purpose is that the interface
`ModelVariable`	Class to represent a variable attached to the model
`GeneVariable`	Class to represent a gene variable
`BinaryVariable`	Class to represent a generic binary variable
`IntegerVariable`	Class to represent a generic binary variable
`ReactionVariable`	Class to represent a variable attached to a reaction
`MetaboliteVariable`	Class to represent a variable attached to a enzyme
`ForwardUseVariable`	Class to represent a forward use variable, a type of binary variable used to
`BackwardUseVariable`	Class to represent a backward use variable, a type of binary variable used
`ForwardBackwardUseVariable`	Class to represent a type of binary variable used to tell whether the
`LogConcentration`	Class to represent a log concentration of a enzyme
`DeltaGErr`	Class to represent a DeltaGErr
`DeltaG`	Class to represent a DeltaG
`ThermoPotential`	Class to represent a DeltaG
`DeltaGFormstd`	Class to represent a DeltaGstf of formation
`DeltaGstd`	Class to represent a DeltaG^o (naught) - standard conditions
`ThermoDisplacement`	Class to represent the thermodynamic displacement of a reaction
`PosSlackVariable`	Class to represent a positive slack variable for relaxation problems
`NegSlackVariable`	Class to represent a negative slack variable for relaxation problems
`PosSlackVariableInt`	Class to represent a postive slack variable for relaxation problems using integer
`NegSlackVariableInt`	Class to represent a postive slack variable for relaxation problems using integer
`PosSlackLC`	Class to represent a variable attached to a enzyme
`NegSlackLC`	Class to represent a variable attached to a enzyme
`LinearizationVariable`	Class to represent a product A*B when performing linearization of the
`NegativeDeltaG`	Class to represent thermodynamics constraints.
`PosSlackVariable`	Class to represent a positive slack variable for relaxation problems
`NegSlackVariable`	Class to represent a negative slack variable for relaxation problems
`DeltaGstd`	Class to represent a DeltaG^o (naught) - standard conditions
`LogConcentration`	Class to represent a log concentration of a enzyme
`NegSlackLC`	Class to represent a variable attached to a enzyme
`PosSlackLC`	Class to represent a variable attached to a enzyme
`PosSlackVariableInt`	Class to represent a postive slack variable for relaxation problems using integer
`NegSlackVariableInt`	Class to represent a postive slack variable for relaxation problems using integer
`GenericConstraint`	Class to represent a generic constraint. The purpose is that the interface
`ForwardUseVariable`	Class to represent a forward use variable, a type of binary variable used to
`BackwardUseVariable`	Class to represent a backward use variable, a type of binary variable used
`GenericVariable`	Class to represent a generic variable. The purpose is that the interface

Functions¶

`camel2underscores`(name)
`camel2underscores`(name)
`get_binary_type`()	FIX : We enforce type to be integer instead of binary, else optlang does
`dg_relax_config`(model)	param model
`get_solution_value_for_variables`(solution, these_vars, index_by_reaction=False)
`chunk_sum`(variables)	This functions handles the sum of many sympy variables by chunks, which
`symbol_sum`(variables)	``` python
`relax_dgo_gurobi`(model, relax_obj_type=0)
`relax_dgo`(tmodel, reactions_to_ignore=(), solver=None, in_place=False)	param t_tmodel
`relax_dgo_err`(tmodel, reactions_to_ignore=(), max_sigma=3, solver=None, in_place=False)	param t_tmodel
`relax_lc`(tmodel, metabolites_to_ignore=(), solver=None)	param metabolites_to_ignore
`get_all_subclasses`(cls)	Given a variable or constraint class, get all the subclassses
`chunk_sum`(variables)	This functions handles the sum of many sympy variables by chunks, which
`symbol_sum`(variables)	``` python
`get_solution_value_for_variables`(solution, these_vars, index_by_reaction=False)
`compare_solutions`(models)	returns the solution dictionnary for each cobra_model
`evaluate_constraint_at_solution`(constraint, solution)	param expression
`get_active_use_variables`(tmodel, solution)	Returns the active use variables in a solution. Use Variables are binary
`get_direction_use_variables`(tmodel, solution)	Returns the active use variables in a solution. Use Variables are binary
`get_primal`(tmodel, vartype, index_by_reactions=False)	Returns the primal value of the cobra_model for variables of a given type
`strip_from_integer_variables`(tmodel)	Removes all integer and binary variables of a cobra_model, to make it sample-able
`copy_solver_configuration`(source, target)	Copies the solver configuration from a source model to a target model

Attributes¶

`ABCRequirePrefixMeta`
`ABCRequirePrefixMeta`
`op_replace_dict`
`BIGM`
`BIGM_THERMO`
`BIGM_DG`
`BIGM_P`
`EPSILON`
`SYMPY_ADD_CHUNKSIZE`
`INTEGER_VARIABLE_TYPES`

pytfa.optim.camel2underscores(name)¶

pytfa.optim.ABCRequirePrefixMeta¶

class pytfa.optim.GenericConstraint(expr, id_='', model=None, hook=None, queue=False, **kwargs)[source]¶

Class to represent a generic constraint. The purpose is that the interface: is instantiated on initialization, to follow the type of interface used by the problem, and avoid incompatibilities in optlang

Attributes:

id

Used for DictList comprehension. Usually points back at a

enzyme or reaction id for ease of linking. Should be unique given a constraint type. :name: Should be a concatenation of the id and a prefix that is specific to the variable type. will be used to address the constraint at the solver level, and hence should be unique in the whole cobra_model :expr: the expression of the constraint (sympy.Expression subtype) :cobra_model: the cobra_model hook. :constraint: links directly to the cobra_model representation of tbe constraint

prefix¶

property __attrname__(self)¶: Name the attribute the instances will have Example: GenericConstraint -> generic_constraint :return:

get_interface(self, expr, queue)¶

Called upon completion of __init__, initializes the value of self.var, which is returned upon call, and stores the actual interfaced variable.

Returns: instance of Variable from the problem

make_name(self)¶

Needs to be overridden by the subclass, concats the id with a: prefix

Returns: None

change_expr(self, new_expr, sloppy=False)¶

property expr(self)¶

property name(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property constraint(self)¶

property model(self)¶

__repr__(self)¶: Return repr(self).

class pytfa.optim.ModelConstraint(model, expr, id_, **kwargs)[source]¶

Bases: GenericConstraint

Class to represent a variable attached to the model

prefix = MODC_¶

class pytfa.optim.GeneConstraint(gene, expr, **kwargs)[source]¶

Bases: GenericConstraint

Class to represent a variable attached to a enzyme

prefix = GC_¶

property gene(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.optim.ReactionConstraint(reaction, expr, **kwargs)[source]¶

Bases: GenericConstraint

Class to represent a variable attached to a reaction

prefix = RC_¶

property reaction(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.optim.MetaboliteConstraint(metabolite, expr, **kwargs)[source]¶

Bases: GenericConstraint

Class to represent a variable attached to a enzyme

prefix = MC_¶

property metabolite(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.optim.NegativeDeltaG(reaction, expr, **kwargs)[source]¶

Bases: ReactionConstraint

Class to represent thermodynamics constraints.

G: - DGR_rxn + DGoRerr_Rxn + RT * StoichCoefProd1 * LC_prod1

RT * StoichCoefProd2 * LC_prod2

RT * StoichCoefSub1 * LC_subs1

RT * StoichCoefSub2 * LC_subs2

…

= 0

prefix = G_¶

class pytfa.optim.NegativeDeltaGIneq(reaction, expr, **kwargs)[source]¶

Bases: ReactionConstraint

Class to represent thermodynamics constraints.

G: DGoRerr_Rxn + RT * StoichCoefProd1 * LC_prod1

RT * StoichCoefProd2 * LC_prod2

RT * StoichCoefSub1 * LC_subs1

RT * StoichCoefSub2 * LC_subs2

…

< or > 0

prefix = GI_¶

class pytfa.optim.ForwardDeltaGCoupling(reaction, expr, **kwargs)[source]¶

Bases: ReactionConstraint

Class to represent thermodynamics coupling: DeltaG of reactions has to be DGR < 0 for the reaction to proceed forwards Looks like: FU_rxn: 1000 FU_rxn + DGR_rxn < 1000

prefix = FU_¶

class pytfa.optim.BackwardDeltaGCoupling(reaction, expr, **kwargs)[source]¶

Bases: ReactionConstraint

Class to represent thermodynamics coupling: DeltaG of reactions has to be DGR > 0 for the reaction to proceed backwards Looks like: BU_rxn: 1000 BU_rxn - DGR_rxn < 1000

prefix = BU_¶

class pytfa.optim.ForwardDirectionCoupling(reaction, expr, **kwargs)[source]¶

Bases: ReactionConstraint

Class to represent a forward directionality coupling with thermodynamics on reaction variables Looks like : UF_rxn: F_rxn - M FU_rxn < 0

prefix = UF_¶

class pytfa.optim.BackwardDirectionCoupling(reaction, expr, **kwargs)[source]¶

Bases: ReactionConstraint

Class to represent a backward directionality coupling with thermodynamics on reaction variables Looks like : UR_rxn: R_rxn - M RU_rxn < 0

prefix = UR_¶

class pytfa.optim.SimultaneousUse(reaction, expr, **kwargs)[source]¶

Bases: ReactionConstraint

Class to represent a simultaneous use constraint on reaction variables Looks like: SU_rxn: FU_rxn + BU_rxn <= 1

prefix = SU_¶

class pytfa.optim.DisplacementCoupling(reaction, expr, **kwargs)[source]¶

Bases: ReactionConstraint

Class to represent the coupling to the thermodynamic displacement Looks like: Ln(Gamma) - (1/RT)*DGR_rxn = 0

prefix = DC_¶

class pytfa.optim.ForbiddenProfile(model, expr, id_, **kwargs)[source]¶

Bases: GenericConstraint

Class to represent a forbidden net flux directionality profile Looks like: FU_rxn_1 + BU_rxn_2 + … + FU_rxn_n <= n-1

prefix = FP_¶

class pytfa.optim.LinearizationConstraint(model, expr, id_, **kwargs)[source]¶

Bases: ModelConstraint

Class to represent a variable attached to a reaction

prefix = LC_¶

static from_constraints(cons, model)¶

class pytfa.optim.LowerBoundLogConcentration(metabolite, expr, **kwargs)[source]¶

Bases: MetaboliteConstraint

Class to represent a forbidden net flux directionality profile Looks like: DG_rxn_1 + DG_rxn_2 + … + DG_rxn_n == 0

prefix = LB_¶

class pytfa.optim.UpperBoundLogConcentration(metabolite, expr, **kwargs)[source]¶

Bases: MetaboliteConstraint

Class to represent a forbidden net flux directionality profile Looks like: DG_rxn_1 + DG_rxn_2 + … + DG_rxn_n == 0

prefix = UB_¶

class pytfa.optim.NullspaceConstraint(metabolite, expr, **kwargs)[source]¶

Bases: MetaboliteConstraint

Class to represent a forbidden net flux directionality profile Looks like: DG_rxn_1 + DG_rxn_2 + … + DG_rxn_n == 0

prefix = NS_¶

class pytfa.optim.PotentialConstraint(metabolite, expr, **kwargs)[source]¶

Bases: MetaboliteConstraint

Class to represent a forbidden net flux directionality profile Looks like: DG_rxn_1 + DG_rxn_2 + … + DG_rxn_n == 0

prefix = PC_¶

class pytfa.optim.PotentialCoupling(reaction, expr, **kwargs)[source]¶

Bases: ReactionConstraint

Class to rep couple DG to potential variable

prefix = PPC_¶

pytfa.optim.camel2underscores(name)¶

pytfa.optim.ABCRequirePrefixMeta¶

pytfa.optim.op_replace_dict[source]¶

class pytfa.optim.GenericVariable(id_='', model=None, hook=None, queue=False, scaling_factor=1, **kwargs)[source]¶

Class to represent a generic variable. The purpose is that the interface is instantiated on initialization, to follow the type of interface used by the problem, and avoid incompatibilities in optlang

Attributes:

id

Used for DictList comprehension. Usually points back at a

enzyme or reaction id for ease of linking. Should be unique given a variable type. :name: Should be a concatenation of the id and a prefix that is specific to the variable type. will be used to address the variable at the solver level, and hence should be unique in the whole cobra_model :cobra_model: the cobra_model hook. :variable: links directly to the cobra_model representation of tbe variable

prefix¶

property __attrname__(self)¶: Name the attribute the instances will have Example: GenericVariable -> generic_variable :return:

get_interface(self, queue)¶

Called upon completion of __init__, initializes the value of self.var, which is returned upon call, and stores the actual interfaced variable.

Returns: instance of Variable from the problem

make_name(self)¶

Needs to be overridden by the subclass, concats the id with a: prefix

Returns: None

property name(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property variable(self)¶

property scaling_factor(self)¶

property unscaled(self)¶

If the scaling factor of quantity X is a, it is represented by the variable X_hat = X/a. This returns X = a.X_hat Useful for nondimensionalisation of variables and constraints

Returns: The variable divided by its scaling factor

property value(self)¶

property unscaled_value(self)¶

property model(self)¶

property type(self)¶

test_consistency(self, other)¶

Tests whether a candidate to an operation is of the right type and is from the same problem

Parameters: other – an object
Returns: None

get_operand(self, other)¶

For operations, choose if the operand is a GenericVariable, in which we return its optlang variable, or something else (presumably a numeric) and we let optlang decide what to do

Parameters: other –
Returns

__add__(self, other)¶: Adding either two variables together or a variable and a numeric results in a new variable :param other: :return: a new Generic Variable

__radd__(self, other)¶: Take priority on symmetric arithmetic operation :param other: :return:

__sub__(self, other)¶: Substracting either two variables together or a variable and a numeric results in a new variable :param other: :return: a new Generic Variable

__rsub__(self, other)¶: Take priority on symmetric arithmetic operation :param other: :return:

__mul__(self, other)¶: Multiplying either two variables together or a variable and a numeric results in a new variable :param other: :return: a new Generic Variable

__rmul__(self, other)¶: Take priority on symmetric arithmetic operation :param other: :return:

__truediv__(self, other)¶: Dividing either two variables together or a variable and a numeric results in a new variable :param other: :return: a new Generic Variable

__rtruediv__(self, other)¶: Take priority on symmetric arithmetic operation :param other: :return:

make_result(self, new_variable)¶: Returns a Sympy expression :param new_variable: :return:

__repr__(self)¶: Return repr(self).

pytfa.optim.get_binary_type()[source]¶: FIX : We enforce type to be integer instead of binary, else optlang does not allow to set the binary variable bounds to anything other than (0,1) You might want to set it at (0,0) to enforce directionality for example

class pytfa.optim.ModelVariable(model, id_, **kwargs)[source]¶

Bases: GenericVariable

Class to represent a variable attached to the model

prefix = MODV_¶

class pytfa.optim.GeneVariable(gene, **kwargs)[source]¶

Bases: GenericVariable

Class to represent a gene variable

prefix = GV_¶

property gene(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.optim.BinaryVariable(id_, model, **kwargs)[source]¶

Bases: GenericVariable

Class to represent a generic binary variable

prefix = B_¶

class pytfa.optim.IntegerVariable(id_, model, **kwargs)[source]¶

Bases: GenericVariable

Class to represent a generic binary variable

prefix = I_¶

class pytfa.optim.ReactionVariable(reaction, **kwargs)[source]¶

Bases: GenericVariable

Class to represent a variable attached to a reaction

prefix = RV_¶

property reaction(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.optim.MetaboliteVariable(metabolite, **kwargs)[source]¶

Bases: GenericVariable

Class to represent a variable attached to a enzyme

prefix = MV_¶

property metabolite(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.optim.ForwardUseVariable(reaction, **kwargs)[source]¶

Bases: ReactionVariable, BinaryVariable

Class to represent a forward use variable, a type of binary variable used to enforce forward directionality in reaction net fluxes

prefix = FU_¶

class pytfa.optim.BackwardUseVariable(reaction, **kwargs)[source]¶

Bases: ReactionVariable, BinaryVariable

Class to represent a backward use variable, a type of binary variable used to enforce backward directionality in reaction net fluxes

prefix = BU_¶

class pytfa.optim.ForwardBackwardUseVariable(reaction, **kwargs)[source]¶

Bases: ReactionVariable, BinaryVariable

Class to represent a type of binary variable used to tell whether the reaction is active or not such that:

FU + BU + BFUSE = 1

prefix = BFUSE_¶

class pytfa.optim.LogConcentration(metabolite, **kwargs)[source]¶

Bases: MetaboliteVariable

Class to represent a log concentration of a enzyme

prefix = LC_¶

class pytfa.optim.DeltaGErr(reaction, **kwargs)[source]¶

Bases: ReactionVariable

Class to represent a DeltaGErr

prefix = DGE_¶

class pytfa.optim.DeltaG(reaction, **kwargs)[source]¶

Bases: ReactionVariable

Class to represent a DeltaG

prefix = DG_¶

class pytfa.optim.ThermoPotential(metabolite, **kwargs)[source]¶

Bases: MetaboliteVariable

Class to represent a DeltaG

prefix = P_¶

class pytfa.optim.DeltaGFormstd(metabolite, **kwargs)[source]¶

Bases: MetaboliteVariable

Class to represent a DeltaGstf of formation

prefix = DGoF_¶

class pytfa.optim.DeltaGstd(reaction, **kwargs)[source]¶

Bases: ReactionVariable

Class to represent a DeltaG^o (naught) - standard conditions

prefix = DGo_¶

class pytfa.optim.ThermoDisplacement(reaction, **kwargs)[source]¶

Bases: ReactionVariable

Class to represent the thermodynamic displacement of a reaction Gamma = -DeltaG/RT

prefix = LnGamma_¶

class pytfa.optim.PosSlackVariable(reaction, **kwargs)[source]¶

Bases: ReactionVariable

Class to represent a positive slack variable for relaxation problems

prefix = PosSlack_¶

class pytfa.optim.NegSlackVariable(reaction, **kwargs)[source]¶

Bases: ReactionVariable

Class to represent a negative slack variable for relaxation problems

prefix = NegSlack_¶

class pytfa.optim.PosSlackVariableInt(reaction, **kwargs)[source]¶

Bases: ReactionVariable, IntegerVariable

Class to represent a postive slack variable for relaxation problems using integer enforce forward directionality in reaction net fluxes

prefix = PosSlackInt_¶

class pytfa.optim.NegSlackVariableInt(reaction, **kwargs)[source]¶

Bases: ReactionVariable, IntegerVariable

Class to represent a postive slack variable for relaxation problems using integer enforce forward directionality in reaction net fluxes

prefix = NegSlackInt_¶

class pytfa.optim.PosSlackLC(metabolite, **kwargs)[source]¶

Bases: MetaboliteVariable

Class to represent a variable attached to a enzyme

prefix = PosSlackLC_¶

class pytfa.optim.NegSlackLC(metabolite, **kwargs)[source]¶

Bases: MetaboliteVariable

Class to represent a variable attached to a enzyme

prefix = NegSlackLC_¶

class pytfa.optim.LinearizationVariable(model, id_, **kwargs)[source]¶

Bases: ModelVariable

Class to represent a product A*B when performing linearization of the model

prefix = LZ_¶

class pytfa.optim.NegativeDeltaG(reaction, expr, **kwargs)[source]¶

Bases: ReactionConstraint

Class to represent thermodynamics constraints.

G: - DGR_rxn + DGoRerr_Rxn + RT * StoichCoefProd1 * LC_prod1

RT * StoichCoefProd2 * LC_prod2

RT * StoichCoefSub1 * LC_subs1

RT * StoichCoefSub2 * LC_subs2

…

= 0

prefix = G_¶

pytfa.optim.dg_relax_config(model)¶

Parameters: model –
Returns

pytfa.optim.get_solution_value_for_variables(solution, these_vars, index_by_reaction=False)¶

pytfa.optim.chunk_sum(variables)¶

This functions handles the sum of many sympy variables by chunks, which somehow increases the speed of the computation

You can test it in IPython: ```python a = sympy.symbols(‘a0:100’) %timeit (sum(a)) # >>> 198 µs ± 11.4 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)

b = sympy.symbols(‘b0:1000’) %timeit (sum(b)) # >>> 1.85 ms ± 356 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)

c = sympy.symbols(‘c0:3000’) %timeit (sum(c)) # >>> 5min 7s ± 2.57 s per loop (mean ± std. dev. of 7 runs, 1 loop each) ```

See the github thread

Parameters: variables –
Returns

pytfa.optim.symbol_sum(variables)¶

``` python a = symbols(‘a0:100’)

%timeit Add(*a) # >>> 10000 loops, best of 3: 34.1 µs per loop

b = symbols(‘b0:1000’)

%timeit Add(*b) # >>> 1000 loops, best of 3: 343 µs per loop

c = symbols(‘c0:3000’)

%timeit Add(*c) # >>> 1 loops, best of 3: 1.03 ms per loop ```

See the github thread :param variables: :return:

class pytfa.optim.PosSlackVariable(reaction, **kwargs)[source]¶

Bases: ReactionVariable

Class to represent a positive slack variable for relaxation problems

prefix = PosSlack_¶

class pytfa.optim.NegSlackVariable(reaction, **kwargs)[source]¶

Bases: ReactionVariable

Class to represent a negative slack variable for relaxation problems

prefix = NegSlack_¶

class pytfa.optim.DeltaGstd(reaction, **kwargs)[source]¶

Bases: ReactionVariable

Class to represent a DeltaG^o (naught) - standard conditions

prefix = DGo_¶

class pytfa.optim.LogConcentration(metabolite, **kwargs)[source]¶

Bases: MetaboliteVariable

Class to represent a log concentration of a enzyme

prefix = LC_¶

class pytfa.optim.NegSlackLC(metabolite, **kwargs)[source]¶

Bases: MetaboliteVariable

Class to represent a variable attached to a enzyme

prefix = NegSlackLC_¶

class pytfa.optim.PosSlackLC(metabolite, **kwargs)[source]¶

Bases: MetaboliteVariable

Class to represent a variable attached to a enzyme

prefix = PosSlackLC_¶

class pytfa.optim.PosSlackVariableInt(reaction, **kwargs)[source]¶

Bases: ReactionVariable, IntegerVariable

Class to represent a postive slack variable for relaxation problems using integer enforce forward directionality in reaction net fluxes

prefix = PosSlackInt_¶

class pytfa.optim.NegSlackVariableInt(reaction, **kwargs)[source]¶

Bases: ReactionVariable, IntegerVariable

Class to represent a postive slack variable for relaxation problems using integer enforce forward directionality in reaction net fluxes

prefix = NegSlackInt_¶

pytfa.optim.BIGM[source]¶

pytfa.optim.BIGM_THERMO[source]¶

pytfa.optim.BIGM_DG[source]¶

pytfa.optim.BIGM_P[source]¶

pytfa.optim.EPSILON[source]¶

pytfa.optim.relax_dgo_gurobi(model, relax_obj_type=0)[source]¶

pytfa.optim.relax_dgo(tmodel, reactions_to_ignore=(), solver=None, in_place=False)[source]¶

Parameters

t_tmodel (pytfa.thermo.ThermoModel:) –
reactions_to_ignore – Iterable of reactions that should not be relaxed
solver – solver to use (e.g. ‘optlang-glpk’, ‘optlang-cplex’, ‘optlang-gurobi’

Returns

a cobra_model with relaxed bounds on standard Gibbs free energy

pytfa.optim.relax_dgo_err(tmodel, reactions_to_ignore=(), max_sigma=3, solver=None, in_place=False)[source]¶

Parameters

t_tmodel (pytfa.thermo.ThermoModel:) –
reactions_to_ignore – Iterable of reactions that should not be relaxed
solver – solver to use (e.g. ‘optlang-glpk’, ‘optlang-cplex’, ‘optlang-gurobi’

Returns

a cobra_model with relaxed bounds on standard Gibbs free energy

pytfa.optim.relax_lc(tmodel, metabolites_to_ignore=(), solver=None)[source]¶

Parameters

metabolites_to_ignore –
in_tmodel (pytfa.thermo.ThermoModel:) –
min_objective_value –

Returns

class pytfa.optim.GenericConstraint(expr, id_='', model=None, hook=None, queue=False, **kwargs)[source]¶

Class to represent a generic constraint. The purpose is that the interface: is instantiated on initialization, to follow the type of interface used by the problem, and avoid incompatibilities in optlang

Attributes:

id

Used for DictList comprehension. Usually points back at a

enzyme or reaction id for ease of linking. Should be unique given a constraint type. :name: Should be a concatenation of the id and a prefix that is specific to the variable type. will be used to address the constraint at the solver level, and hence should be unique in the whole cobra_model :expr: the expression of the constraint (sympy.Expression subtype) :cobra_model: the cobra_model hook. :constraint: links directly to the cobra_model representation of tbe constraint

prefix¶

property __attrname__(self)¶: Name the attribute the instances will have Example: GenericConstraint -> generic_constraint :return:

get_interface(self, expr, queue)¶

Called upon completion of __init__, initializes the value of self.var, which is returned upon call, and stores the actual interfaced variable.

Returns: instance of Variable from the problem

make_name(self)¶

Needs to be overridden by the subclass, concats the id with a: prefix

Returns: None

change_expr(self, new_expr, sloppy=False)¶

property expr(self)¶

property name(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property constraint(self)¶

property model(self)¶

__repr__(self)¶: Return repr(self).

class pytfa.optim.ForwardUseVariable(reaction, **kwargs)[source]¶

Bases: ReactionVariable, BinaryVariable

Class to represent a forward use variable, a type of binary variable used to enforce forward directionality in reaction net fluxes

prefix = FU_¶

class pytfa.optim.BackwardUseVariable(reaction, **kwargs)[source]¶

Bases: ReactionVariable, BinaryVariable

Class to represent a backward use variable, a type of binary variable used to enforce backward directionality in reaction net fluxes

prefix = BU_¶

class pytfa.optim.GenericVariable(id_='', model=None, hook=None, queue=False, scaling_factor=1, **kwargs)[source]¶

Class to represent a generic variable. The purpose is that the interface is instantiated on initialization, to follow the type of interface used by the problem, and avoid incompatibilities in optlang

Attributes:

id

Used for DictList comprehension. Usually points back at a

enzyme or reaction id for ease of linking. Should be unique given a variable type. :name: Should be a concatenation of the id and a prefix that is specific to the variable type. will be used to address the variable at the solver level, and hence should be unique in the whole cobra_model :cobra_model: the cobra_model hook. :variable: links directly to the cobra_model representation of tbe variable

prefix¶

property __attrname__(self)¶: Name the attribute the instances will have Example: GenericVariable -> generic_variable :return:

get_interface(self, queue)¶

Called upon completion of __init__, initializes the value of self.var, which is returned upon call, and stores the actual interfaced variable.

Returns: instance of Variable from the problem

make_name(self)¶

Needs to be overridden by the subclass, concats the id with a: prefix

Returns: None

property name(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property variable(self)¶

property scaling_factor(self)¶

property unscaled(self)¶

If the scaling factor of quantity X is a, it is represented by the variable X_hat = X/a. This returns X = a.X_hat Useful for nondimensionalisation of variables and constraints

Returns: The variable divided by its scaling factor

property value(self)¶

property unscaled_value(self)¶

property model(self)¶

property type(self)¶

test_consistency(self, other)¶

Tests whether a candidate to an operation is of the right type and is from the same problem

Parameters: other – an object
Returns: None

get_operand(self, other)¶

For operations, choose if the operand is a GenericVariable, in which we return its optlang variable, or something else (presumably a numeric) and we let optlang decide what to do

Parameters: other –
Returns

__add__(self, other)¶: Adding either two variables together or a variable and a numeric results in a new variable :param other: :return: a new Generic Variable

__radd__(self, other)¶: Take priority on symmetric arithmetic operation :param other: :return:

__sub__(self, other)¶: Substracting either two variables together or a variable and a numeric results in a new variable :param other: :return: a new Generic Variable

__rsub__(self, other)¶: Take priority on symmetric arithmetic operation :param other: :return:

__mul__(self, other)¶: Multiplying either two variables together or a variable and a numeric results in a new variable :param other: :return: a new Generic Variable

__rmul__(self, other)¶: Take priority on symmetric arithmetic operation :param other: :return:

__truediv__(self, other)¶: Dividing either two variables together or a variable and a numeric results in a new variable :param other: :return: a new Generic Variable

__rtruediv__(self, other)¶: Take priority on symmetric arithmetic operation :param other: :return:

make_result(self, new_variable)¶: Returns a Sympy expression :param new_variable: :return:

__repr__(self)¶: Return repr(self).

pytfa.optim.SYMPY_ADD_CHUNKSIZE = 100[source]¶

pytfa.optim.INTEGER_VARIABLE_TYPES = ['binary', 'integer'][source]¶

pytfa.optim.get_all_subclasses(cls)[source]¶

Given a variable or constraint class, get all the subclassses that inherit from it

Parameters: cls –
Returns

pytfa.optim.chunk_sum(variables)¶

This functions handles the sum of many sympy variables by chunks, which somehow increases the speed of the computation

You can test it in IPython: ```python a = sympy.symbols(‘a0:100’) %timeit (sum(a)) # >>> 198 µs ± 11.4 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)

b = sympy.symbols(‘b0:1000’) %timeit (sum(b)) # >>> 1.85 ms ± 356 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)

c = sympy.symbols(‘c0:3000’) %timeit (sum(c)) # >>> 5min 7s ± 2.57 s per loop (mean ± std. dev. of 7 runs, 1 loop each) ```

See the github thread

Parameters: variables –
Returns

pytfa.optim.symbol_sum(variables)¶

``` python a = symbols(‘a0:100’)

%timeit Add(*a) # >>> 10000 loops, best of 3: 34.1 µs per loop

b = symbols(‘b0:1000’)

%timeit Add(*b) # >>> 1000 loops, best of 3: 343 µs per loop

c = symbols(‘c0:3000’)

%timeit Add(*c) # >>> 1 loops, best of 3: 1.03 ms per loop ```

See the github thread :param variables: :return:

pytfa.optim.get_solution_value_for_variables(solution, these_vars, index_by_reaction=False)¶

pytfa.optim.compare_solutions(models)[source]¶: returns the solution dictionnary for each cobra_model :param (iterable (pytfa.thermo.ThermoModel)) models: :return:

pytfa.optim.evaluate_constraint_at_solution(constraint, solution)[source]¶

Parameters

expression –
solution – pandas.DataFrame, with index as variable names

Returns

pytfa.optim.get_active_use_variables(tmodel, solution)[source]¶

Returns the active use variables in a solution. Use Variables are binary variables that control the directionality of the reaction

ex: FU_ACALDt BU_PFK

Parameters

tmodel (pytfa.core.ThermoModel) –
solution –

Returns

pytfa.optim.get_direction_use_variables(tmodel, solution)[source]¶

Returns the active use variables in a solution. Use Variables are binary variables that control the directionality of the reaction The difference with get_active_use_variables is that variables with both UseVariables at 0 will return as going forwards. This is to ensure that the output size of the function is equal to the number of FDPs

ex: FU_ACALDt BU_PFK

Parameters

tmodel (pytfa.core.ThermoModel) –
solution –

Returns

pytfa.optim.get_primal(tmodel, vartype, index_by_reactions=False)[source]¶

Returns the primal value of the cobra_model for variables of a given type :param tmodel: :param vartype: Class of variable. Ex: pytfa.optim.variables.ThermoDisplacement :param index_by_reactions: Set to true to get reaction names as index instead of

variables. Useful for Escher export

Returns

pytfa.optim.strip_from_integer_variables(tmodel)[source]¶: Removes all integer and binary variables of a cobra_model, to make it sample-able :param tmodel: :return:

pytfa.optim.copy_solver_configuration(source, target)[source]¶: Copies the solver configuration from a source model to a target model :param source: :param target: :return:

`pytfa.redgem`¶

Submodules¶

pytfa.redgem.debugging¶

Debugging

Module Contents¶

Functions¶

`make_sink`(met, ub=100, lb=0)
`add_BBB_sinks`(model, biomass_rxn_id, ub=100, lb=0)
`check_BBB_production`(model, biomass_rxn_id, verbose=False)
`min_BBB_uptake`(model, biomass_rxn_id, min_growth_value, verbose=False)

redgem.make_sink(met, ub=100, lb=0)¶

redgem.add_BBB_sinks(model, biomass_rxn_id, ub=100, lb=0)¶

redgem.check_BBB_production(model, biomass_rxn_id, verbose=False)¶

redgem.min_BBB_uptake(model, biomass_rxn_id, min_growth_value, verbose=False)¶

pytfa.redgem.lumpgem¶

Module Contents¶

Classes¶

`FluxKO`	Class to represent a variable attached to a reaction
`UseOrKOInt`	Class to represent a variable attached to a reaction
`UseOrKOFlux`	Class to represent a variable attached to a reaction
`LumpGEM`	A class encapsulating the LumpGEM algorithm

Functions¶

sum_reactions(rxn_dict, id_='summed_reaction', epsilon=1e-09, exclude=())

Keys are reactions

Attributes¶

`CPLEX`
`GUROBI`
`GLPK`
`DEFAULT_EPS`
`disambiguate`
`Lump`

pytfa.redgem.lumpgem.CPLEX = optlang-cplex[source]¶

pytfa.redgem.lumpgem.GUROBI = optlang-gurobi[source]¶

pytfa.redgem.lumpgem.GLPK = optlang-glpk[source]¶

pytfa.redgem.lumpgem.DEFAULT_EPS = 1e-05[source]¶

pytfa.redgem.lumpgem.disambiguate[source]¶

pytfa.redgem.lumpgem.Lump[source]¶

exception pytfa.redgem.lumpgem.InfeasibleExcept(status, feasibility)[source]¶

Bases: Exception

Common base class for all non-exit exceptions.

exception pytfa.redgem.lumpgem.TimeoutExcept(time_limit)[source]¶

Bases: Exception

Common base class for all non-exit exceptions.

class pytfa.redgem.lumpgem.FluxKO(reaction, **kwargs)[source]¶

Bases: pytfa.optim.variables.ReactionVariable, pytfa.optim.variables.BinaryVariable

Class to represent a variable attached to a reaction

prefix = KO_[source]¶

class pytfa.redgem.lumpgem.UseOrKOInt(reaction, expr, **kwargs)[source]¶

Bases: pytfa.optim.constraints.ReactionConstraint

Class to represent a variable attached to a reaction

prefix = UKI_[source]¶

class pytfa.redgem.lumpgem.UseOrKOFlux(reaction, expr, **kwargs)[source]¶

Bases: pytfa.optim.constraints.ReactionConstraint

Class to represent a variable attached to a reaction

prefix = UKF_[source]¶

class pytfa.redgem.lumpgem.LumpGEM(tfa_model, additional_core_reactions, params)[source]¶

A class encapsulating the LumpGEM algorithm

init_params(self)[source]¶

_generate_usage_constraints(self)[source]¶: Generate carbon intake related constraints for each non-core reaction For each reaction rxn : rxn.forward_variable + rxn.reverse_variable + activation_var * C_uptake < C_uptake

get_cofactor_adjusted_stoich(self, rxn)[source]¶

_prepare_sinks(self)[source]¶: For each BBB (reactant of the biomass reactions), generate a sink, i.e an unbalanced reaction BBB -> of which purpose is to enable the BBB to be output of the GEM :return: the dict {BBB: sink} containing every BBB (keys) and their associated sinks

_generate_objective(self)[source]¶: Generate and add the maximization objective : set as many activation variables as possible to 1 When an activation variable is set to 1, the corresponding non-core reaction is deactivated

compute_lumps(self, force_solve=False, method='OnePerBBB')[source]¶: For each BBB (reactant of the biomass reaction), add the corresponding sink to the model, then optimize and lump the result into one lumped reaction :param force_solve: Indicates whether the computations must continue when one lumping yields a status “infeasible” :return: The dict {BBB: lump} containing every lumped reactions, associated to their BBBs

_lump_one_per_bbb(self, met_BBB, sink, force_solve, minimize_subnet_fluxes=False)[source]¶

Parameters

met_BBB –
sink –
force_solve –

Returns

_lump_min_plus_p(self, met_BBB, sink, p, force_solve)[source]¶

Parameters

met_BBB –
sink –
force_solve –

Returns

_build_lump(self, met_BBB, sink)[source]¶

This function uses the current solution of self._tfa_model

Parameters

met_BBB –
sink –

Returns

pytfa.redgem.lumpgem.sum_reactions(rxn_dict, id_='summed_reaction', epsilon=1e-09, exclude=())[source]¶: Keys are reactions Values are their multiplicative coefficient

pytfa.redgem.network_expansion¶

Model class

Module Contents¶

Classes¶

NetworkExpansion

class redgem.NetworkExpansion(gem, core_subsystems, extracellular_system, cofactors, small_metabolites, inorganics, d, n)¶

extract_subsystem_reactions(self, subsystem)¶

Extracts all reactions of a subsystem and stores them and their id in the corresponding dictionary.

Parameters: subsystem – Name of the subsystem
Returns: Extracted reactions

extract_subsystem_metabolites(self, subsystem)¶

Extracts all metabolites of a subsystem and stores them and their id in the corresponding dictionary.

Parameters: subsystem – Name of the subsystem
Returns: Extracted metabolites

create_new_stoichiometric_matrix(self)¶

Extracts the new graph without the small metabolites, inorganics and cofactor pairs.

Returns: Networkx graph of the new network

breadth_search_subsystems_paths_length_d(self, subsystem_i, subsystem_j, d)¶

Breadth first search from each metabolite in subsystem i with special stop conditions during exploration for paths of length d.

This function explores the graph through allowed paths only : this path can’t go through subsystem i or j but must start in i and end in j. The length of each path found is d.

Parameters

subsystem_i – Source subsystem
subsystem_j – Destination subsystem
d – Path length desired

Returns

None

is_node_allowed(self, node, i, explored, subsystem_i, subsystem_j, d)¶

Checks whether or not a metabolite is allowed for the current path.

The new node is added if it is not already explored, if it is not in the source subsystem, and if it is not in the destination subsystem, except if it is the last round of exploration

Parameters

node – Metabolite id
i – Current step
explored – Explored node for this path
subsystem_i – Source subsystem
subsystem_j – Destination subsystem
d – Path length desired

Returns

Boolean answering the question

retrieve_all_paths(self, dest_node, src_node, ancestors, init_dict=True)¶

Retrieves all paths between a source metabolite and a destination metabolite after a breadth first search.

This function is a recursive function, which makes use of dynamic programming to reduce its complexity. It uses self._path_dict to store already computed data.

Parameters

dest_node – Destination metabolite
src_node – Source metabolite
ancestors – Dictionary with ancestors found during the search
init_dict – Boolean, for function initialisation

Returns

A list of all paths as tuples

retrieve_intermediate_metabolites_and_reactions(self, paths, subsystem_i, subsystem_j, d)¶

Retrieves and stores intermediate metabolites and reactions (i.e. M_{i,j}, R_{i,j}, M_{i,i} and R_{i,i}).

This function adds all reactions contained in these paths, and all metabolites between

Parameters

paths – List of paths between subsystems
subsystem_i – Source subsystem
subsystem_j – Destination subsystem
d – Path length

Returns

None

find_min_distance_between_subsystems(self)¶

Find minimal distance between each subsystems in both directions

Returns: Dict with distances

breadth_search_extracellular_system_paths(self, subsystem, n)¶

Breadth first search from each metabolite in the extracellular system with special stop conditions during exploration for paths of length n.

This function explores the graph through allowed paths only : this path can’t go through the extracellular system or the subsystem but must start in the extracellular system and end in the subsystem. The length of each path found is n.

Parameters

subsystem – Destination subsystem
n – Path length desired

Returns

None

is_node_allowed_extracellular(self, node, i, explored, subsystem, n)¶

Checks whether or not a metabolite is allowed for the current path.

The new node is added if it is not already explored, if it is not in the extracellular system, and if it is not in the destination subsystem except if it is the last round of exploration

Parameters

node – Metabolite id
i – Current step
explored – Explored node for this path
subsystem – Destination subsystem
n – Path length desired

Returns

Boolean answering the question

retrieve_intermediate_extracellular_metabolites_and_reactions(self, paths, subsystem, n)¶

Retrieves and stores intermediate metabolites and reactions for the extracellular system

This function adds all reactions contained in these paths, and all metabolites between

Parameters

paths – List of paths
subsystem – Destination subsystem
n – Path length

Returns

None

run_between_all_subsystems(self)¶

Retrieve subsystem and intermediate reactions and metabolites.

Returns: None

run_extracellular_system(self)¶

Retrieve intermediate reactions and metabolites for the extracellular system

Returns: None

extract_sub_network(self, additional_reactions)¶

Extracts the reduced gem.

Returns: None

run(self, additional_reactions=[])¶

Runs RedGEM.

Returns: None

pytfa.redgem.redgem¶

Model class

Module Contents¶

Classes¶

RedGEM

Functions¶

add_lump(model, lump_object, id_suffix='')

class redgem.RedGEM(gem, parameters_path, inplace=False)¶

read_parameters(self, parameters_path)¶

fill_default_params(self)¶

set_solver(self)¶

run(self)¶

_extract_inorganics(self)¶

Extract inorganics from self._gem based on their formula

Returns: list of inorganics metabolites

redgem.add_lump(model, lump_object, id_suffix='')¶

pytfa.redgem.utils¶

Module Contents¶

Functions¶

`remove_blocked_reactions`(model)
`round`(value, epsilon)
`trim_epsilon_mets`(met_dict, epsilon, model=None)
`set_medium`(model, medium_dict, inplace)

pytfa.redgem.utils.remove_blocked_reactions(model)[source]¶

pytfa.redgem.utils.round(value, epsilon)[source]¶

pytfa.redgem.utils.trim_epsilon_mets(met_dict, epsilon, model=None)[source]¶

pytfa.redgem.utils.set_medium(model, medium_dict, inplace)[source]¶

`pytfa.thermo`¶

Thermodynamic analysis for Flux-Based Analysis

Submodules¶

pytfa.thermo.equilibrator¶

Thermodynamic information for metabolites from eQuilibrator.

Module Contents¶

Functions¶

`build_thermo_from_equilibrator`(model, T=TEMPERATURE_0)	Build thermo_data structure from a cobra Model.
`compute_dGf`(compound, cc)	Get ΔGf from equilibrator compound.
`compound_to_entry`(compound, cc)	Build thermo structure entry from a equilibrator_cache.Compound.

Attributes¶

`ccache`
`logger`

pytfa.ccache¶

pytfa.logger¶

pytfa.build_thermo_from_equilibrator(model, T=TEMPERATURE_0)¶

Build thermo_data structure from a cobra Model.

The structure of the returned dictionary is specified in the pyTFA [documentation](https://pytfa.readthedocs.io/en/latest/thermoDB.html).

Parameters: model – cobra.Model
Return thermo_data: dict to be passed as argument to initialize a ThermoModel.

pytfa.compute_dGf(compound, cc)¶: Get ΔGf from equilibrator compound.

pytfa.compound_to_entry(compound, cc)¶

Build thermo structure entry from a equilibrator_cache.Compound.

eQuilibrator works with Component Contribution instead of groups, so it is not possible to generate cues from it.

Parameters: compound – equilibrator_cache.Compound
Returns: dict with keys [‘deltaGf_std’, ‘deltaGf_err’, ‘error’, ‘struct_cues’, ‘id’, ‘pKa’, ‘mass_std’, ‘charge_std’, ‘nH_std’, ‘name’, ‘formula’, ‘other_names’]

pytfa.thermo.metabolite¶

Thermodynamic computations for metabolites

Module Contents¶

Classes¶

MetaboliteThermo

A class representing the thermodynamic values of a enzyme

Attributes¶

`CPD_PROTON`
`DEFAULT_VAL`

pytfa.CPD_PROTON = cpd00067¶

pytfa.DEFAULT_VAL¶

class pytfa.MetaboliteThermo(metData, pH, ionicStr, temperature=std.TEMPERATURE_0, min_ph=std.MIN_PH, max_ph=std.MAX_PH, debye_huckel_b=std.DEBYE_HUCKEL_B_0, thermo_unit='kJ/mol', debug=False)¶

A class representing the thermodynamic values of a enzyme

Parameters

metData (dict) – A dictionary containing the values for the enzyme, from the thermodynamic database
pH (float) – The pH of the enzyme’s compartment
ionicStr (float) – The ionic strength of the enzyme’s compartment
temperature –
min_ph –
max_ph –
debye_huckel_b –
thermo_unit (string) – The unit used in metData’s values
debug (bool) – Optional If set to True, some debugging values will be printed. This is only useful for development or debugging purposes.

Note

The values are automatically computed on class creation. Usually you don’t have to call any methods defined by this class, but only to access the attributes you need.

The available attributes are :

Since the reactions expose similar values through a dictionnary, it is better to access the attributes aforementionned of this class as if it was a dictionnary : enzyme.thermo['pH'].

__getitem__(self, key)¶

__repr__(self)¶: Return repr(self).

keys(self)¶

values(self)¶

items(self)¶

__cmp__(self, dict_)¶

__contains__(self, item)¶

__iter__(self)¶

__unicode__(self)¶

calcDGis(self)¶

Calculate the transformed Gibbs energy of formation of specie with given pH and ionic strength using formula given by Goldberg and Tewari, 1991

Equation 4.5-6 in Alberty’s book

Returns: DG_is for the enzyme
Return type: float

calcDGsp(self)¶

Calculate the transformed Gibbs energy of formation of specie with given pH and ionic strength using formula given by Goldberg and Tewari, 1991

Equation 4.4-10 in Alberty’s book

Returns: DG_sp for the enzyme
Return type: float

calc_potential(self)¶

Calculate the binding polynomial of a specie, with pK values

Returns: The potential of the enzyme
Return type: float

get_pka(self)¶

Get the pKas of the enzyme

Returns: The pKas of the enzyme
Return type: list(float)

_calc_pka(self, pka, sigmanusq)¶

calcDGspA(self)¶

Calculates deltaGf, charge and nH of the specie when it is at least protonated state based on MFAToolkit compound data for the pKa values within the range considered (MIN_pH to MAX_pH).

These values are used as the starting point for Alberty’s calculations.

Returns: deltaGspA, sp_charge and sp_nH
Return type: tuple(float, float, int)

pytfa.thermo.reaction¶

Thermodynamic computations for reactions

Module Contents¶

Functions¶

`calcDGtpt_rhs`(reaction, compartmentsData, thermo_units)	Calculates the RHS of the deltaG constraint, i.e. the sum of the
`calcDGR_cues`(reaction, reaction_cues_data)	Calculates the deltaG reaction and error of the reaction using the
`calcDGF_cues`(cues, reaction_cues_data)	Calculates the deltaG formation and error of the compound using its
`get_debye_huckel_b`(T)	The Debye-Huckel A and B do depend on the temperature

pytfa.calcDGtpt_rhs(reaction, compartmentsData, thermo_units)¶

Calculates the RHS of the deltaG constraint, i.e. the sum of the non-concentration terms

Parameters

reaction (cobra.thermo.reaction.Reaction) – The reaction to compute the data for
compartmentsData (dict(float)) – Data of the compartments of the cobra_model
thermo_units (str) – The thermodynamic database of the cobra_model

Returns

deltaG_tpt and the breakdown of deltaG_tpt

Return type

tuple(float, dict(float))

Example:

ATP Synthase reaction:

reaction = cpd00008 + 4 cpd00067 + cpd00009 <=> cpd00002 + 3 cpd00067 + cpd00001
compartments =  'c'       'e'        'c'           'c'         'c'         'c'

If there are any metabolites with unknown energies then returns

(0, None).

pytfa.calcDGR_cues(reaction, reaction_cues_data)¶

Calculates the deltaG reaction and error of the reaction using the constituent structural cues changes and returns also the error if any.

Parameters

reaction (cobra.thermo.reaction.Reaction) – The reaction to compute deltaG for
reaction_cues_data (dict) –

Returns

deltaGR, error on deltaGR, the cues in the reaction (keys of the dictionnary) and their indices (values of the dictionnary), and the error code if any.

If everything went right, the error code is an empty string

Return type

tuple(float, float, dict(float), str)

pytfa.calcDGF_cues(cues, reaction_cues_data)¶

Calculates the deltaG formation and error of the compound using its constituent structural cues.

Parameters

cues (list(str)) – A list of cues’ names
reaction_cues_data (dict) –

Returns

deltaG formation, the error on deltaG formation, and a dictionnary with the cues’ names as key and their coefficient as value

Return type

tuple(float, float, dict(float))

pytfa.get_debye_huckel_b(T)¶

The Debye-Huckel A and B do depend on the temperature As for now though they are returned as a constant (value at 298.15K)

Parameters: T – Temperature in Kelvin
Returns: Debye_Huckel_B

pytfa.thermo.std¶

Standard constants definitions

Module Contents¶

pytfa.thermo.std.TEMPERATURE_0 = 298.15[source]¶

pytfa.thermo.std.MIN_PH = 3[source]¶

pytfa.thermo.std.MAX_PH = 9[source]¶

pytfa.thermo.std.DEBYE_HUCKEL_B_0 = 1.6[source]¶

pytfa.thermo.std.DEBYE_HUCKEL_A[source]¶

pytfa.thermo.std.A_LOT = 5000[source]¶

pytfa.thermo.std.A_LITTLE = 0.5[source]¶

pytfa.thermo.std.A_COUPLE = 2.5[source]¶

pytfa.thermo.std.MANY = 100[source]¶

pytfa.thermo.tmodel¶

Thermodynamic cobra_model class and methods definition

Module Contents¶

Classes¶

ThermoModel

A class representing a cobra_model with thermodynamics information

Attributes¶

`BIGM`
`BIGM_THERMO`
`BIGM_DG`
`BIGM_P`
`EPSILON`
`MAX_STOICH`

pytfa.BIGM¶

pytfa.BIGM_THERMO¶

pytfa.BIGM_DG¶

pytfa.BIGM_P¶

pytfa.EPSILON¶

pytfa.MAX_STOICH = 10¶

class pytfa.ThermoModel(thermo_data=None, model=Model(), name=None, temperature=std.TEMPERATURE_0, min_ph=std.MIN_PH, max_ph=std.MAX_PH)[source]¶

Bases: pytfa.core.model.LCSBModel, cobra.Model

A class representing a cobra_model with thermodynamics information

_init_thermo(self)¶

normalize_reactions(self)¶: Find reactions with important stoichiometry and normalizes them :return:

_prepare_metabolite(self, met)¶

Parameters: met –
Returns

_prepare_reaction(self, reaction)¶

prepare(self)¶

Prepares a COBRA toolbox cobra_model for TFBA analysis by doing the following:

checks if a reaction is a transport reaction
checks the ReactionDB for Gibbs energies of formation of metabolites
computes the Gibbs energies of reactions

_convert_metabolite(self, met, add_potentials, verbose)¶

Given a enzyme, proceeds to create the necessary variables and constraints for thermodynamics-based modeling

Parameters: met –
Returns

_convert_reaction(self, rxn, add_potentials, add_displacement, verbose)¶

Parameters

rxn –
add_potentials –
add_displacement –
verbose –

Returns

convert(self, add_potentials=False, add_displacement=False, verbose=True)¶: Converts a cobra_model into a tFBA ready cobra_model by adding the thermodynamic constraints required

Warning

This function requires you to have already called prepare(), otherwise it will raise an Exception !

print_info(self, specific=False)¶: Print information and counts for the cobra_model :return:

__deepcopy__(self, memo)¶

Parameters: memo –
Returns

copy(self)¶: Needs to be reimplemented, as our objects have complicated hierarchy :return:

pytfa.thermo.utils¶

Some tools around COBRApy models used by pyTFA

Module Contents¶

Functions¶

`check_reaction_balance`(reaction, proton=None)	Check the balance of a reaction, and eventually add protons to balance
`find_transported_mets`(reaction)	Get a list of the transported metabolites of the reaction.
`check_transport_reaction`(reaction)	Check if a reaction is a transport reaction
`is_same_stoichiometry`(this_reaction, that_reaction)
`is_exchange`(rxn)
`get_reaction_compartment`(reaction)	Get the compartment of a reaction to then prepare it for conversion.

Attributes¶

Formula_regex

pytfa.Formula_regex¶

pytfa.check_reaction_balance(reaction, proton=None)¶

Check the balance of a reaction, and eventually add protons to balance it

Parameters

reaction (cobra.thermo.reaction.Reaction) – The reaction to check the balance of.
proton (cobra.thermo.metabolite.Metabolite) – Optional The proton to add to the reaction to balance it.

Returns

The balance of the reaction :

drain flux
missing structures
balanced
N protons added to reactants with N a float
N protons added to products with N a float
missing atoms

Return type

str

If proton is provided, this function will try to balance the equation with it, and return the result.

If no proton is provided, this function will not try to balance the equation.

Warning

This function does not verify if proton is in the correct compartment, so make sure you provide the proton belonging to the correct compartment !

pytfa.find_transported_mets(reaction)¶

Get a list of the transported metabolites of the reaction.

Parameters

reaction (cobra.thermo.reaction.Reaction) – The reaction to get the transported metabolites of

Returns

A dictionnary of the transported metabolites. The index corresponds to the seed_id of the transported enzyme

The value is a dictionnairy with the following values:

coeff (float):
The stoechiomectric coefficient of the enzyme

reactant (cobra.thermo.enzyme.Metabolite):
The reactant of the reaction corresponding to the transported enzyme

product (cobra.thermo.enzyme.Metabolite):
The product of the reaction corresponding to the transported enzyme

A transported enzyme is defined as a enzyme which is a product and a reactant of a reaction. We can distinguish them thanks to their seed_ids.

pytfa.check_transport_reaction(reaction)¶

Check if a reaction is a transport reaction

Parameters: reaction (cobra.thermo.reaction.Reaction) – The reaction to check
Returns: Whether the reaction is a transport reaction or not
Return type: bool

A transport reaction is defined as a reaction that has the same compound as a reactant and a product. We can distinguish them thanks to their seed_ids. If they have one If not, use met_ids and check if they are the same, minus compartment

pytfa.is_same_stoichiometry(this_reaction, that_reaction)¶

pytfa.is_exchange(rxn)¶

pytfa.get_reaction_compartment(reaction)¶: Get the compartment of a reaction to then prepare it for conversion.

Package Contents¶

Classes¶

`LCSBModel`	Helper class that provides a standard way to create an ABC using
`MetaboliteThermo`	A class representing the thermodynamic values of a enzyme
`SimultaneousUse`	Class to represent a simultaneous use constraint on reaction variables
`NegativeDeltaG`	Class to represent thermodynamics constraints.
`BackwardDeltaGCoupling`	Class to represent thermodynamics coupling: DeltaG of reactions has to be
`ForwardDeltaGCoupling`	Class to represent thermodynamics coupling: DeltaG of reactions has to be
`BackwardDirectionCoupling`	Class to represent a backward directionality coupling with thermodynamics on
`ForwardDirectionCoupling`	Class to represent a forward directionality coupling with thermodynamics on
`ReactionConstraint`	Class to represent a variable attached to a reaction
`MetaboliteConstraint`	Class to represent a variable attached to a enzyme
`DisplacementCoupling`	Class to represent the coupling to the thermodynamic displacement
`PotentialConstraint`	Class to represent a forbidden net flux directionality profile
`PotentialCoupling`	Class to rep couple DG to potential variable
`ThermoDisplacement`	Class to represent the thermodynamic displacement of a reaction
`DeltaGstd`	Class to represent a DeltaG^o (naught) - standard conditions
`DeltaG`	Class to represent a DeltaG
`ForwardUseVariable`	Class to represent a forward use variable, a type of binary variable used to
`BackwardUseVariable`	Class to represent a backward use variable, a type of binary variable used
`LogConcentration`	Class to represent a log concentration of a enzyme
`ReactionVariable`	Class to represent a variable attached to a reaction
`MetaboliteVariable`	Class to represent a variable attached to a enzyme
`ThermoPotential`	Class to represent a DeltaG
`DeltaGFormstd`	Class to represent a DeltaGstf of formation
`ThermoModel`	A class representing a cobra_model with thermodynamics information
`MetaboliteThermo`	A class representing the thermodynamic values of a enzyme

Functions¶

`calcDGtpt_rhs`(reaction, compartmentsData, thermo_units)	Calculates the RHS of the deltaG constraint, i.e. the sum of the
`calcDGR_cues`(reaction, reaction_cues_data)	Calculates the deltaG reaction and error of the reaction using the
`get_debye_huckel_b`(T)	The Debye-Huckel A and B do depend on the temperature
`check_reaction_balance`(reaction, proton=None)	Check the balance of a reaction, and eventually add protons to balance
`check_transport_reaction`(reaction)	Check if a reaction is a transport reaction
`find_transported_mets`(reaction)	Get a list of the transported metabolites of the reaction.
`get_bistream_logger`(name)	Sets up a logger that outputs INFO+ messages on stdout and DEBUG+ messages

Attributes¶

`BIGM`
`BIGM_THERMO`
`BIGM_DG`
`BIGM_P`
`EPSILON`
`MAX_STOICH`

class pytfa.thermo.LCSBModel(model, name, sloppy=False)¶

Bases: abc.ABC

Helper class that provides a standard way to create an ABC using inheritance.

abstract copy(self)¶: Needs to be reimplemented, as our objects have complicated hierarchy :return:

print_info(self)¶: Print information and counts for the cobra_model :return:

add_variable(self, kind, hook, queue=False, **kwargs)¶

Add a new variable to a COBRApy cobra_model.

Parameters

kind –
hook (string,cobra.Reaction) – Either a string representing the name of the variable to add to the cobra_model, or a reaction object if the kind allows it

Returns

The created variable

Return type

optlang.interface.Variable

add_constraint(self, kind, hook, expr, queue=False, **kwargs)¶

Add a new constraint to a COBRApy cobra_model

Parameters

kind –
hook (string,cobra.Reaction) – Either a string representing the name of the variable to add to the cobra_model, or a reaction object if the kind allows it
expr (sympy.thermo.expr.Expr) – The expression of the constraint

Returns

The created constraint

Return type

optlang.interface.Constraint

remove_reactions(self, reactions, remove_orphans=False)¶

remove_metabolites(self, metabolite_list, destructive=False)¶

_remove_associated_consvar(self, all_cons_subclasses, all_var_subclasses, collection)¶: Removes both the constraints and variables associated to an element, as long as it was used as a hook in the cons/var declaration. For example, upon removing a reaction, also removes its associated deltaG variables and coupling constraints

remove_variable(self, var)¶

Removes a variable

Parameters: var –
Returns

remove_constraint(self, cons)¶

Removes a constraint

Parameters: cons –
Returns

_push_queue(self)¶: updates the constraints and variables of the model with what’s in the queue :return:

regenerate_variables(self)¶: Generates references to the cobra_model’s constraints in self._var_dict as tab-searchable attributes of the thermo cobra_model :return:

regenerate_constraints(self)¶: Generates references to the cobra_model’s constraints in self._cons_dict as tab-searchable attributes of the thermo cobra_model :return:

repair(self)¶: Updates references to variables and constraints :return:

get_primal(self, vartype, index_by_reactions=False)¶

Returns the primal value of the cobra_model for variables of a given type

Parameters

index_by_reactions –
vartype – Class of variable. Ex: pytfa.optim.variables.ThermoDisplacement

Returns

get_solution(self)¶

Overrides the cobra.thermo.solution method, to also get the supplementary variables we added to the cobra_model

solution.fluxes in cobrapy is a transformed version of the solver output, as it actually calculates the _net_ flux of each reaction by substracting the reverse variable value to the forward variable value. This should be used anytime one needs the actual flux value
solution.raw is a clear copy of the solver output. From there one can access the value at solution for all the variables of the problem. However, looking for a reaction ID in there will only give the _forward_ flux. This should be used for any other variable than fluxes.
solution.values yields variables multiplied by their scaling factor (1 by default). Useful if you operated scaling on your equations for numerical reasons. This does _not_ include fluxes

Returns

optimize(self, objective_sense=None, **kwargs)¶: Call the Model.optimize function (which really is but an interface to the solver’s. Catches SolverError in the case of no solutions. Passes down supplementary keyword arguments (see cobra.thermo.Model.optimize) :type objective_sense: ‘min’ or ‘max’

slim_optimize(self, *args, **kwargs)¶

get_constraints_of_type(self, constraint_type)¶

Convenience function that takes as input a constraint class and returns all its instances within the cobra_model

Parameters: constraint_type –
Returns

get_variables_of_type(self, variable_type)¶

Convenience function that takes as input a variable class and returns all its instances within the cobra_model

Parameters: variable_type –
Returns

class pytfa.thermo.MetaboliteThermo(metData, pH, ionicStr, temperature=std.TEMPERATURE_0, min_ph=std.MIN_PH, max_ph=std.MAX_PH, debye_huckel_b=std.DEBYE_HUCKEL_B_0, thermo_unit='kJ/mol', debug=False)¶

A class representing the thermodynamic values of a enzyme

Parameters

metData (dict) – A dictionary containing the values for the enzyme, from the thermodynamic database
pH (float) – The pH of the enzyme’s compartment
ionicStr (float) – The ionic strength of the enzyme’s compartment
temperature –
min_ph –
max_ph –
debye_huckel_b –
thermo_unit (string) – The unit used in metData’s values
debug (bool) – Optional If set to True, some debugging values will be printed. This is only useful for development or debugging purposes.

Note

The values are automatically computed on class creation. Usually you don’t have to call any methods defined by this class, but only to access the attributes you need.

The available attributes are :

Since the reactions expose similar values through a dictionnary, it is better to access the attributes aforementionned of this class as if it was a dictionnary : enzyme.thermo['pH'].

__getitem__(self, key)¶

__repr__(self)¶: Return repr(self).

keys(self)¶

values(self)¶

items(self)¶

__cmp__(self, dict_)¶

__contains__(self, item)¶

__iter__(self)¶

__unicode__(self)¶

calcDGis(self)¶

Calculate the transformed Gibbs energy of formation of specie with given pH and ionic strength using formula given by Goldberg and Tewari, 1991

Equation 4.5-6 in Alberty’s book

Returns: DG_is for the enzyme
Return type: float

calcDGsp(self)¶

Calculate the transformed Gibbs energy of formation of specie with given pH and ionic strength using formula given by Goldberg and Tewari, 1991

Equation 4.4-10 in Alberty’s book

Returns: DG_sp for the enzyme
Return type: float

calc_potential(self)¶

Calculate the binding polynomial of a specie, with pK values

Returns: The potential of the enzyme
Return type: float

get_pka(self)¶

Get the pKas of the enzyme

Returns: The pKas of the enzyme
Return type: list(float)

_calc_pka(self, pka, sigmanusq)¶

calcDGspA(self)¶

Calculates deltaGf, charge and nH of the specie when it is at least protonated state based on MFAToolkit compound data for the pKa values within the range considered (MIN_pH to MAX_pH).

These values are used as the starting point for Alberty’s calculations.

Returns: deltaGspA, sp_charge and sp_nH
Return type: tuple(float, float, int)

pytfa.thermo.calcDGtpt_rhs(reaction, compartmentsData, thermo_units)¶

Calculates the RHS of the deltaG constraint, i.e. the sum of the non-concentration terms

Parameters

reaction (cobra.thermo.reaction.Reaction) – The reaction to compute the data for
compartmentsData (dict(float)) – Data of the compartments of the cobra_model
thermo_units (str) – The thermodynamic database of the cobra_model

Returns

deltaG_tpt and the breakdown of deltaG_tpt

Return type

tuple(float, dict(float))

Example:

ATP Synthase reaction:

reaction = cpd00008 + 4 cpd00067 + cpd00009 <=> cpd00002 + 3 cpd00067 + cpd00001
compartments =  'c'       'e'        'c'           'c'         'c'         'c'

If there are any metabolites with unknown energies then returns

(0, None).

pytfa.thermo.calcDGR_cues(reaction, reaction_cues_data)¶

Calculates the deltaG reaction and error of the reaction using the constituent structural cues changes and returns also the error if any.

Parameters

reaction (cobra.thermo.reaction.Reaction) – The reaction to compute deltaG for
reaction_cues_data (dict) –

Returns

deltaGR, error on deltaGR, the cues in the reaction (keys of the dictionnary) and their indices (values of the dictionnary), and the error code if any.

If everything went right, the error code is an empty string

Return type

tuple(float, float, dict(float), str)

pytfa.thermo.get_debye_huckel_b(T)¶

The Debye-Huckel A and B do depend on the temperature As for now though they are returned as a constant (value at 298.15K)

Parameters: T – Temperature in Kelvin
Returns: Debye_Huckel_B

pytfa.thermo.check_reaction_balance(reaction, proton=None)¶

Check the balance of a reaction, and eventually add protons to balance it

Parameters

reaction (cobra.thermo.reaction.Reaction) – The reaction to check the balance of.
proton (cobra.thermo.metabolite.Metabolite) – Optional The proton to add to the reaction to balance it.

Returns

The balance of the reaction :

drain flux
missing structures
balanced
N protons added to reactants with N a float
N protons added to products with N a float
missing atoms

Return type

str

If proton is provided, this function will try to balance the equation with it, and return the result.

If no proton is provided, this function will not try to balance the equation.

Warning

This function does not verify if proton is in the correct compartment, so make sure you provide the proton belonging to the correct compartment !

pytfa.thermo.check_transport_reaction(reaction)¶

Check if a reaction is a transport reaction

Parameters: reaction (cobra.thermo.reaction.Reaction) – The reaction to check
Returns: Whether the reaction is a transport reaction or not
Return type: bool

A transport reaction is defined as a reaction that has the same compound as a reactant and a product. We can distinguish them thanks to their seed_ids. If they have one If not, use met_ids and check if they are the same, minus compartment

pytfa.thermo.find_transported_mets(reaction)¶

Get a list of the transported metabolites of the reaction.

Parameters

reaction (cobra.thermo.reaction.Reaction) – The reaction to get the transported metabolites of

Returns

A dictionnary of the transported metabolites. The index corresponds to the seed_id of the transported enzyme

The value is a dictionnairy with the following values:

coeff (float):
The stoechiomectric coefficient of the enzyme

reactant (cobra.thermo.enzyme.Metabolite):
The reactant of the reaction corresponding to the transported enzyme

product (cobra.thermo.enzyme.Metabolite):
The product of the reaction corresponding to the transported enzyme

A transported enzyme is defined as a enzyme which is a product and a reactant of a reaction. We can distinguish them thanks to their seed_ids.

class pytfa.thermo.SimultaneousUse(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent a simultaneous use constraint on reaction variables Looks like: SU_rxn: FU_rxn + BU_rxn <= 1

prefix = SU_¶

class pytfa.thermo.NegativeDeltaG(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent thermodynamics constraints.

G: - DGR_rxn + DGoRerr_Rxn + RT * StoichCoefProd1 * LC_prod1

RT * StoichCoefProd2 * LC_prod2

RT * StoichCoefSub1 * LC_subs1

RT * StoichCoefSub2 * LC_subs2

…

= 0

prefix = G_¶

class pytfa.thermo.BackwardDeltaGCoupling(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent thermodynamics coupling: DeltaG of reactions has to be DGR > 0 for the reaction to proceed backwards Looks like: BU_rxn: 1000 BU_rxn - DGR_rxn < 1000

prefix = BU_¶

class pytfa.thermo.ForwardDeltaGCoupling(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent thermodynamics coupling: DeltaG of reactions has to be DGR < 0 for the reaction to proceed forwards Looks like: FU_rxn: 1000 FU_rxn + DGR_rxn < 1000

prefix = FU_¶

class pytfa.thermo.BackwardDirectionCoupling(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent a backward directionality coupling with thermodynamics on reaction variables Looks like : UR_rxn: R_rxn - M RU_rxn < 0

prefix = UR_¶

class pytfa.thermo.ForwardDirectionCoupling(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent a forward directionality coupling with thermodynamics on reaction variables Looks like : UF_rxn: F_rxn - M FU_rxn < 0

prefix = UF_¶

class pytfa.thermo.ReactionConstraint(reaction, expr, **kwargs)¶

Bases: GenericConstraint

Class to represent a variable attached to a reaction

prefix = RC_¶

property reaction(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.thermo.MetaboliteConstraint(metabolite, expr, **kwargs)¶

Bases: GenericConstraint

Class to represent a variable attached to a enzyme

prefix = MC_¶

property metabolite(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.thermo.DisplacementCoupling(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to represent the coupling to the thermodynamic displacement Looks like: Ln(Gamma) - (1/RT)*DGR_rxn = 0

prefix = DC_¶

class pytfa.thermo.PotentialConstraint(metabolite, expr, **kwargs)¶

Bases: MetaboliteConstraint

Class to represent a forbidden net flux directionality profile Looks like: DG_rxn_1 + DG_rxn_2 + … + DG_rxn_n == 0

prefix = PC_¶

class pytfa.thermo.PotentialCoupling(reaction, expr, **kwargs)¶

Bases: ReactionConstraint

Class to rep couple DG to potential variable

prefix = PPC_¶

class pytfa.thermo.ThermoDisplacement(reaction, **kwargs)¶

Bases: ReactionVariable

Class to represent the thermodynamic displacement of a reaction Gamma = -DeltaG/RT

prefix = LnGamma_¶

class pytfa.thermo.DeltaGstd(reaction, **kwargs)¶

Bases: ReactionVariable

Class to represent a DeltaG^o (naught) - standard conditions

prefix = DGo_¶

class pytfa.thermo.DeltaG(reaction, **kwargs)¶

Bases: ReactionVariable

Class to represent a DeltaG

prefix = DG_¶

class pytfa.thermo.ForwardUseVariable(reaction, **kwargs)¶

Bases: ReactionVariable, BinaryVariable

Class to represent a forward use variable, a type of binary variable used to enforce forward directionality in reaction net fluxes

prefix = FU_¶

class pytfa.thermo.BackwardUseVariable(reaction, **kwargs)¶

Bases: ReactionVariable, BinaryVariable

Class to represent a backward use variable, a type of binary variable used to enforce backward directionality in reaction net fluxes

prefix = BU_¶

class pytfa.thermo.LogConcentration(metabolite, **kwargs)¶

Bases: MetaboliteVariable

Class to represent a log concentration of a enzyme

prefix = LC_¶

class pytfa.thermo.ReactionVariable(reaction, **kwargs)¶

Bases: GenericVariable

Class to represent a variable attached to a reaction

prefix = RV_¶

property reaction(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.thermo.MetaboliteVariable(metabolite, **kwargs)¶

Bases: GenericVariable

Class to represent a variable attached to a enzyme

prefix = MV_¶

property metabolite(self)¶

property id(self)¶: for cobra.thermo.DictList compatibility :return:

property model(self)¶

class pytfa.thermo.ThermoPotential(metabolite, **kwargs)¶

Bases: MetaboliteVariable

Class to represent a DeltaG

prefix = P_¶

class pytfa.thermo.DeltaGFormstd(metabolite, **kwargs)¶

Bases: MetaboliteVariable

Class to represent a DeltaGstf of formation

prefix = DGoF_¶

pytfa.thermo.get_bistream_logger(name)¶

Sets up a logger that outputs INFO+ messages on stdout and DEBUG+ messages in the log file

Parameters: name – a class __name__ attribute
Returns

pytfa.thermo.BIGM[source]¶

pytfa.thermo.BIGM_THERMO[source]¶

pytfa.thermo.BIGM_DG[source]¶

pytfa.thermo.BIGM_P[source]¶

pytfa.thermo.EPSILON[source]¶

pytfa.thermo.MAX_STOICH = 10[source]¶

class pytfa.thermo.ThermoModel(thermo_data=None, model=Model(), name=None, temperature=std.TEMPERATURE_0, min_ph=std.MIN_PH, max_ph=std.MAX_PH)[source]¶

Bases: pytfa.core.model.LCSBModel, cobra.Model

A class representing a cobra_model with thermodynamics information

_init_thermo(self)¶

normalize_reactions(self)¶: Find reactions with important stoichiometry and normalizes them :return:

_prepare_metabolite(self, met)¶

Parameters: met –
Returns

_prepare_reaction(self, reaction)¶

prepare(self)¶

Prepares a COBRA toolbox cobra_model for TFBA analysis by doing the following:

checks if a reaction is a transport reaction
checks the ReactionDB for Gibbs energies of formation of metabolites
computes the Gibbs energies of reactions

_convert_metabolite(self, met, add_potentials, verbose)¶

Given a enzyme, proceeds to create the necessary variables and constraints for thermodynamics-based modeling

Parameters: met –
Returns

_convert_reaction(self, rxn, add_potentials, add_displacement, verbose)¶

Parameters

rxn –
add_potentials –
add_displacement –
verbose –

Returns

convert(self, add_potentials=False, add_displacement=False, verbose=True)¶: Converts a cobra_model into a tFBA ready cobra_model by adding the thermodynamic constraints required

Warning

This function requires you to have already called prepare(), otherwise it will raise an Exception !

print_info(self, specific=False)¶: Print information and counts for the cobra_model :return:

__deepcopy__(self, memo)¶

Parameters: memo –
Returns

copy(self)¶: Needs to be reimplemented, as our objects have complicated hierarchy :return:

class pytfa.thermo.MetaboliteThermo(metData, pH, ionicStr, temperature=std.TEMPERATURE_0, min_ph=std.MIN_PH, max_ph=std.MAX_PH, debye_huckel_b=std.DEBYE_HUCKEL_B_0, thermo_unit='kJ/mol', debug=False)¶

A class representing the thermodynamic values of a enzyme

Parameters

metData (dict) – A dictionary containing the values for the enzyme, from the thermodynamic database
pH (float) – The pH of the enzyme’s compartment
ionicStr (float) – The ionic strength of the enzyme’s compartment
temperature –
min_ph –
max_ph –
debye_huckel_b –
thermo_unit (string) – The unit used in metData’s values
debug (bool) – Optional If set to True, some debugging values will be printed. This is only useful for development or debugging purposes.

Note

The values are automatically computed on class creation. Usually you don’t have to call any methods defined by this class, but only to access the attributes you need.

The available attributes are :

Since the reactions expose similar values through a dictionnary, it is better to access the attributes aforementionned of this class as if it was a dictionnary : enzyme.thermo['pH'].

__getitem__(self, key)¶

__repr__(self)¶: Return repr(self).

keys(self)¶

values(self)¶

items(self)¶

__cmp__(self, dict_)¶

__contains__(self, item)¶

__iter__(self)¶

__unicode__(self)¶

calcDGis(self)¶

Calculate the transformed Gibbs energy of formation of specie with given pH and ionic strength using formula given by Goldberg and Tewari, 1991

Equation 4.5-6 in Alberty’s book

Returns: DG_is for the enzyme
Return type: float

calcDGsp(self)¶

Calculate the transformed Gibbs energy of formation of specie with given pH and ionic strength using formula given by Goldberg and Tewari, 1991

Equation 4.4-10 in Alberty’s book

Returns: DG_sp for the enzyme
Return type: float

calc_potential(self)¶

Calculate the binding polynomial of a specie, with pK values

Returns: The potential of the enzyme
Return type: float

get_pka(self)¶

Get the pKas of the enzyme

Returns: The pKas of the enzyme
Return type: list(float)

_calc_pka(self, pka, sigmanusq)¶

calcDGspA(self)¶

Calculates deltaGf, charge and nH of the specie when it is at least protonated state based on MFAToolkit compound data for the pKa values within the range considered (MIN_pH to MAX_pH).

These values are used as the starting point for Alberty’s calculations.

Returns: deltaGspA, sp_charge and sp_nH
Return type: tuple(float, float, int)

`pytfa.utils`¶

Submodules¶

pytfa.utils.logger¶

Logging utilities

Module Contents¶

Functions¶

`get_bistream_logger`(name)	Sets up a logger that outputs INFO+ messages on stdout and DEBUG+ messages
`get_timestr`()

Attributes¶

LOGFOLDERNAME

pytfa.LOGFOLDERNAME = logs¶

pytfa.get_bistream_logger(name)¶

Sets up a logger that outputs INFO+ messages on stdout and DEBUG+ messages in the log file

Parameters: name – a class __name__ attribute
Returns

pytfa.get_timestr()¶

pytfa.utils.numerics¶

BIG M and epsilon constants definitions

Module Contents¶

pytfa.utils.numerics.BIGM = 1000[source]¶

pytfa.utils.numerics.BIGM_THERMO = 1000.0[source]¶

pytfa.utils.numerics.BIGM_DG = 1000.0[source]¶

pytfa.utils.numerics.BIGM_P = 1000.0[source]¶

pytfa.utils.numerics.EPSILON = 1e-06[source]¶

pytfa.utils.str¶

Some tools used by pyTFA

Module Contents¶

Functions¶

`camel2underscores`(name)
`varnames2ids`(tmodel, variables)

pytfa.camel2underscores(name)¶

pytfa.varnames2ids(tmodel, variables)¶

Package Contents¶

Classes¶

ThermoModel

A class representing a cobra_model with thermodynamics information

class pytfa.ThermoModel(thermo_data=None, model=Model(), name=None, temperature=std.TEMPERATURE_0, min_ph=std.MIN_PH, max_ph=std.MAX_PH)[source]¶

Bases: pytfa.core.model.LCSBModel, cobra.Model

A class representing a cobra_model with thermodynamics information

_init_thermo(self)¶

normalize_reactions(self)¶: Find reactions with important stoichiometry and normalizes them :return:

_prepare_metabolite(self, met)¶

Parameters: met –
Returns

_prepare_reaction(self, reaction)¶

prepare(self)¶

Prepares a COBRA toolbox cobra_model for TFBA analysis by doing the following:

checks if a reaction is a transport reaction
checks the ReactionDB for Gibbs energies of formation of metabolites
computes the Gibbs energies of reactions

_convert_metabolite(self, met, add_potentials, verbose)¶

Given a enzyme, proceeds to create the necessary variables and constraints for thermodynamics-based modeling

Parameters: met –
Returns

_convert_reaction(self, rxn, add_potentials, add_displacement, verbose)¶

Parameters

rxn –
add_potentials –
add_displacement –
verbose –

Returns

convert(self, add_potentials=False, add_displacement=False, verbose=True)¶: Converts a cobra_model into a tFBA ready cobra_model by adding the thermodynamic constraints required

Warning

This function requires you to have already called prepare(), otherwise it will raise an Exception !

print_info(self, specific=False)¶: Print information and counts for the cobra_model :return:

__deepcopy__(self, memo)¶

Parameters: memo –
Returns

copy(self)¶: Needs to be reimplemented, as our objects have complicated hierarchy :return:

1: Created with sphinx-autoapi

Welcome to the documentation of PyTFA !¶

Quick start¶

pyTFA models¶

Compartment data¶

Metabolites¶

Reactions¶

Thermodynamic Databases¶

Converting a Matlab thermodynamic database¶

Loading a thermodynamic database¶

Structure of a thermodynamic database¶

Metabolites¶

Cues¶

Solver setup¶

GPLK¶

CPLEX¶

Gurobi¶

Integrating metabolomics¶

API Reference¶

Subpackages¶

Submodules¶

Package Contents¶

Submodules¶

Package Contents¶

Submodules¶

Package Contents¶

Submodules¶

Package Contents¶

Submodules¶

Submodules¶

Package Contents¶

Submodules¶

Package Contents¶

Classes¶

Indices and tables¶