fit.petab_v2.reader¶

Read a PEtab v2 problem into an sbmlsim optimization problem.

The tables of a PEtab problem are translated into a SimulationExperiment, in the same way a SED-ML parser builds an experiment from a SED-ML document: the models of the problem are the models of the experiment, its experiments are the timecourse simulations, its observables are the fit mappings and the measurements of an observable are its dataset.

A problem which was written by sbmlsim.fit.petab_v2.export carries the sbmlsim extension, which holds what the tables do not: the units, the settings of the fit, the kind of every mapping and the timecourses with their output grid. The reader uses it when it is there, so that a round trip gives the fit which was written, and falls back on the tables when it is not, which is the case for a problem of another tool.

PetabReader ¶

PetabReader(
    petab_problem,
    base_path=None,
    name=None,
    derived_dir=None,
    sciml=None,
)

Read a PEtab v2 problem as an sbmlsim simulation experiment and fit.

Attributes:

Name	Type	Description
`sciml`	`SciMLReader \| None`	the networks of a problem of PEtab SciML, `None` for a problem without them.

Initialize the reader.

Parameters:

Name	Type	Description	Default
`petab_problem`	`Problem`	problem to read.	required
`base_path`	`Path \| None`	directory the files of the problem are relative to, the directory of its YAML file by default.	`None`
`name`	`str \| None`	name of the simulation experiment class which is created, the id of the problem by default.	`None`
`derived_dir`	`Path \| None`	directory the models the fit simulates are written to, next to the model by default: the model which carries the networks of the problem, `<stem>_sciml.xml`, and the model which carries its observables, `<stem>_observables.xml`. A model is only written if the problem has networks in the model or an observable which is a formula.	`None`
`sciml`	`SciMLConfig \| Mapping[str, Any] \| None`	the block of the extension of PEtab SciML, a `SciMLConfig` or the dictionary of the YAML. It is the block of the configuration of the problem by default, which a problem carries that `petab` has read with its networks. `from_yaml` reads the tables without the block and hands it over.	`None`

Raises:

Type	Description
`ImportError`	if the problem has neural networks and the extra `sciml` is not installed.
`ValueError`	if the problem has no model or no measurements, if it requires an extension `sbmlsim` does not know, or if its networks cannot be read, see `sbmlsim.fit.petab_v2.sciml.SciMLReader`.

settings `property` ¶

settings

Get the settings of the fit, the defaults without the extension.

experiment_ids `property` ¶

experiment_ids

Get the experiments which are simulated.

A measurement which names no experiment is "use the model as is", i.e. DEFAULT_EXPERIMENT, which a problem of one condition does not have to declare (PEtab v2, measurement table).

from_yaml `staticmethod` ¶

from_yaml(yaml_file, name=None)

Read the problem of a PEtab YAML file.

Parameters:

Name	Type	Description	Default
`yaml_file`	`Path`	path of the YAML file of the problem.	required
`name`	`str \| None`	name of the simulation experiment class which is created.	`None`

Returns:

Type	Description
`PetabReader`	The reader of the problem.

Raises:

Type	Description
`ImportError`	if the problem has neural networks and the extra `sciml` is not installed.
`ValueError`	if the problem requires an extension `sbmlsim` does not know. The extensions are checked on the YAML, before `petab` reads the problem: it needs the package of an extension to read its files.

observable_info ¶

observable_info(key)

Get what the extension says about a fit mapping, empty without one.

Parameters:

Name	Type	Description	Default
`key`	`str`	key of the fit mapping. The block `observables` of the extension is keyed by the fit mapping since its version 0.2.0, and by the observable before: an observable measured in several experiments is one observable of several fit mappings, each with its own kind and weight.	required

models ¶

models()

Get the models of the experiment, one per model of the problem.

An observable which is a formula over the entities of the model is not something roadrunner selects, so the model the fit simulates carries it as an entity, see sbmlsim.fit.petab_v2.observables.

model_source ¶

model_source(model_id=None)

Get the file of the model the fit simulates.

Parameters:

Name	Type	Description	Default
`model_id`	`str \| None`	id of the model, the first model of the problem by default.	`None`

Returns:

Type	Description
`Path`	The path of the model, see `_model_source`.

Raises:

Type	Description
`ValueError`	if the problem has no model of the id.

simulations ¶

simulations()

Get the simulations, one per experiment of the problem.

The timecourses of the extension are used if the problem carries it, which keeps the output grid and the pre-simulations of the fit which was written. Without it a timecourse is built from the periods of the experiment and the times of the measurements it holds.

tasks ¶

tasks()

Get the tasks, one per experiment of the problem.

datasets ¶

datasets()

Get the datasets, one per fit mapping, from its measurements.

fit_mappings ¶

fit_mappings(experiment)

Get the fit mappings, one per observable and experiment.

Parameters:

Name	Type	Description	Default
`experiment`	`SimulationExperiment`	experiment the mappings belong to.	required

Returns:

Type	Description
`dict[str, FitMapping]`	The mappings by their key, which is the id of their observable,
`dict[str, FitMapping]`	and `<observable>_<experiment>` for an observable which is
`dict[str, FitMapping]`	measured in several experiments.

observable_id ¶

observable_id(key)

Get the observable of a fit mapping of the problem.

Parameters:

Name	Type	Description	Default
`key`	`str`	key of the fit mapping, which is the id of its observable, and `<observable>_<experiment>` for an observable which is measured in several experiments.	required

Returns:

Type	Description
`str`	The id of the observable.

Raises:

Type	Description
`ValueError`	if the problem has no fit mapping of the key.

noise_model ¶

noise_model(key)

Get the noise model of a fit mapping of the problem.

The noise model is read once, see _read_noise_model.

Parameters:

Name	Type	Description	Default
`key`	`str`	key of the fit mapping, which is the id of its observable, and `<observable>_<experiment>` for an observable which is measured in several experiments.	required

Returns:

Type	Description
`NoiseModel`	The noise model of the fit mapping.

Raises:

Type	Description
`ValueError`	if the problem has no fit mapping of the key, or if a measurement does not have a value for every placeholder.

experiment_class ¶

experiment_class()

Create the simulation experiment class of the problem.

Returns:

Type	Description
`type[SimulationExperiment]`	A `SimulationExperiment` subclass which holds the models, the
`type[SimulationExperiment]`	simulations, the tasks, the datasets and the fit mappings of the
`type[SimulationExperiment]`	PEtab problem.

fit_parameters ¶

fit_parameters()

Get the parameters which are estimated.

A parameter which a condition assigns to an entity of the model, see _versions, is a versioned parameter: it is not itself an entity of a model, so it is exempt from the check which otherwise drops a parameter PEtab estimates that sbmlsim cannot fit, and it is built with the target it writes and the mappings selector of its keys.

Returns:

Type	Description
`list[FitParameter]`	The parameters with their bounds, their start value and, if the
`list[FitParameter]`	problem carries the extension, their unit.

mapping_collections ¶

mapping_collections(experiment_class)

Get the fit mapping collections of the problem, one per experiment.

An experiment of PEtab is a simulation with its conditions, and the observables which are measured in it are the mappings which belong together, i.e. one FitMappingCollection per experiment of the problem. The collection carries the id of the experiment and, with the sbmlsim extension, the kind the mappings had; without it everything is training data, which is what a PEtab problem means.

Parameters:

Name	Type	Description	Default
`experiment_class`	`type[SimulationExperiment]`	the simulation experiment the mappings belong to.	required

Returns:

Type	Description
`list[FitMappingCollection]`	The collections, in the order of the experiments of the problem.

to_optimization_problem ¶

to_optimization_problem(opid=None)

Build the optimization problem of the PEtab problem.

Parameters:

Name	Type	Description	Default
`opid`	`str \| None`	id of the problem, the id of the PEtab problem by default.	`None`

Returns:

Type	Description
`OptimizationProblem`	The problem, which is not initialized: `initialize(settings)` with
`OptimizationProblem`	the `settings` of the reader resolves its data.

dataset_id ¶

dataset_id(observable_id)

Get the id of the dataset which holds the measurements of an observable.

from_petab ¶

from_petab(yaml_file, opid=None)

Read a PEtab v2 problem as an optimization problem.

Parameters:

Name	Type	Description	Default
`yaml_file`	`Path`	path of the YAML file of the PEtab problem.	required
`opid`	`str \| None`	id of the optimization problem.	`None`

Returns:

Type	Description
`OptimizationProblem`	The problem and the settings of the fit, i.e. what `run_optimization`
`FitSettings`	needs. The settings are the defaults if the problem does not carry the
`tuple[OptimizationProblem, FitSettings]`	`sbmlsim` extension.

fit.petab_v2.reader¶

PetabReader ¶

settings property ¶

experiment_ids property ¶

from_yaml staticmethod ¶

observable_info ¶

models ¶

model_source ¶

simulations ¶

tasks ¶

datasets ¶

fit_mappings ¶

observable_id ¶

noise_model ¶

experiment_class ¶

fit_parameters ¶

mapping_collections ¶

to_optimization_problem ¶

dataset_id ¶

from_petab ¶

settings `property` ¶

experiment_ids `property` ¶

from_yaml `staticmethod` ¶