fit.petab_v2.extension¶

The sbmlsim extension of a PEtab v2 problem.

PEtab v2 problems carry extensions, i.e., a block in the YAML of the problem which a tool reads if it knows it and ignores if it does not. The sbmlsim extension holds what the tables of PEtab do not express, see sbmlsim.fit.petab_v2.gaps: the units, the settings of the fit, what a fit does with a mapping and the simulation behind every experiment.

The extension is required, because the settings it carries are the objective of the fit: a tool which does not know it has to reject the problem rather than fit the same data with another objective without saying so. A round trip through sbmlsim keeps the fit it started from, and to_petab(..., required_extension=False) writes a problem which other tools read and fit with the objective PEtab defines.

SbmlsimExtension ¶

Bases: ExtensionConfig

What PEtab v2 does not express about an sbmlsim fit.

PEtab says that an extension which changes the mathematical interpretation of a problem must be required, and that a tool must reject a problem which requires an extension it does not know but may ignore one which is not required (PEtab v2, extensions). The settings this extension carries, i.e. the residual, the loss function, the weighting and the state a measurement at the time of a change is compared with (data_at_change), are the objective sbmlsim optimizes, so a problem which is read without them is fitted with a different objective on the same data. It is therefore required, and a tool which does not know sbmlsim has to say so instead of fitting the problem differently without telling anyone.

to_petab(..., required_extension=False) writes it as not required, which is what a problem meant for other tools wants: they read the tables and optimize the objective PEtab defines.

Attributes:

Name	Type	Description
`version`	`str`	version of the extension.
`required`	`bool`	whether a tool needs the extension to interpret the problem, `True` because the settings it carries are the objective of the fit.
`opid`	`str \| None`	id of the optimization problem.
`settings`	`dict[str, Any]`	the `FitSettings` of the fit as a dictionary.
`parameters`	`dict[str, dict[str, Any]]`	unit, start value and, when set, the scale of every fit parameter which is no element of a network.
`observables`	`dict[str, dict[str, Any]]`	the fit mapping behind every observable, keyed by the fit mapping (an observable measured in several experiments is one observable of several fit mappings): the observable, its kind, the weight of the curve, the units of the data, the experiment and the task it belongs to and the metadata of the curve.
`experiments`	`dict[str, dict[str, Any]]`	the `Simulation` behind every PEtab experiment as its dictionary (`simulation`), i.e. the changes with their units and the output, and the fit mapping collection it belongs to. A problem of a version before 0.3.0 holds the timecourses it is converted from, see `simulation_of_timecourses`.
`collections`	`dict[str, dict[str, Any]]`	the `FitMappingCollection` objects of the fit, i.e. the id of the collection, the simulation experiment class its mappings come from and what the fit does with them.
`models`	`dict[str, dict[str, Any]]`	the settings of the integrator per model.
`inputs`	`dict[str, str]`	the formula of an input of a network for the simulations without a formula of their own, in the math of PEtab, by the id of the input. The conditions of these experiments repeat it, see `sbmlsim.fit.petab_v2.sciml_export`.
`gaps`	`list[dict[str, Any]]`	what this export lost, see `sbmlsim.fit.petab_v2.gaps`.

extension_of ¶

extension_of(config)

Get the sbmlsim extension of a PEtab problem configuration.

Parameters:

Name	Type	Description	Default
`config`	`Any`	`ProblemConfig` of a PEtab v2 problem, `None` if the problem was not read from a YAML file.	required

Returns:

Type	Description
`SbmlsimExtension \| None`	The extension, or `None` if the problem does not carry one.

sciml_installed ¶

sciml_installed()

Check whether the extra sciml is installed, i.e. petab_sciml.

known_extensions ¶

known_extensions()

Get the extensions the reader interprets in this environment.

Returns:

Type	Description
`frozenset[str]`	`KNOWN_EXTENSIONS`, and the extension of PEtab SciML when the extra
`frozenset[str]`	`sciml` is installed.

check_extensions ¶

check_extensions(extensions, known=None)

Check the extensions of a problem against the ones the reader knows.

PEtab says that a tool must reject a problem which requires an extension it does not know and may ignore an extension which is not required (PEtab v2, extensions). A block without required is read as required, which is the safe reading of a block that does not say.

Parameters:

Name	Type	Description	Default
`extensions`	`Mapping[str, Any] \| None`	the blocks of the problem by the id of the extension, as the dictionaries of the YAML or as the `ExtensionConfig` objects of a problem which was read, `None` for a problem without extensions.	required
`known`	`Collection[str] \| None`	ids of the extensions the reader interprets, `known_extensions` by default.	`None`

Returns:

Type	Description
`list[str]`	The ids of the extensions which are to be ignored, i.e. the ones
`list[str]`	which are not known and not required, in the order of the problem.
`list[str]`	The reader logs them.

Raises:

Type	Description
`ImportError`	if the problem requires the extension of PEtab SciML and the extra `sciml` is not installed. The message names the extra.
`ValueError`	if the problem requires an extension which is not known.

simulation_of_timecourses ¶

simulation_of_timecourses(info, ureg)

Convert the timecourses of an extension before 0.3.0 into a Simulation.

The extension stored the TimecourseSim of sbmlsim before 0.9.0: the timecourses with a relative interval each, the time offset of the simulation and the discarded timecourses of a pre-simulation. A timecourse starts where the one before it ended; a leading discarded timecourse runs before the time offset and is the start of the simulation. The changes of the first timecourse are the pre-initialization changes, the ones of a later timecourse a Change at its start, and the output are the grids of the kept timecourses.

Parameters:

Name	Type	Description	Default
`info`	`Mapping[str, Any]`	the block of an experiment, with `timecourses` and `time_offset`.	required
`ureg`	`UnitRegistry`	the registry the units of the changes are read with.	required

Returns:

Type	Description
`Simulation`	The simulation.

Raises:

Type	Description
`ValueError`	if a discarded timecourse follows a kept one.