Skip to content

model.provenance

The record of what was derived from a model.

sbmlsim writes models which are derived from the model of a problem: the model with the compiled networks of a hybrid problem (sbmlsim.sciml.compiler.compile_network) and the model with the observables which are formulas (sbmlsim.fit.petab_v2.observables.add_observables). A fit simulates the derived model, an export writes the model the problem was defined with. The derived model therefore carries what was added to it, as an element of the annotation of the model in the namespace NAMESPACE:

<derived xmlns="https://github.com/matthiaskoenig/sbmlsim/derived" source="lv.xml">
  <created>net1__layer1__weight__0_0 net1__output0__0</created>
  <target id="gamma" constant="true"/>
</derived>

created lists the parameters which were added, with their rules, and target lists the parameters of the source which got a rule, with the value of constant they had. strip_derivation undoes both, which gives the source model. A derivation of a derived model extends the record, so the source is always the model the problem was defined with.

The element is added to the annotation of the model in place: libsbml checks an annotation it is given as a whole against the metaid of the model, and the RDF annotation of a model of the wild does not always have one.

Derivation dataclass 

Derivation(source, created=(), targets=())

What was added to a model.

Attributes:

Name Type Description
source str

name of the file of the model the derivation started from.
created tuple[str, ...]

ids of the parameters which were added, in the order they were added; a rule of such a parameter was added with it.
targets tuple[tuple[str, bool], ...]

the parameters of the source which got a rule, as pairs of the id and whether it was constant before, in the order they got it; dict(derivation.targets) is the lookup. Pairs and not a mapping, so the record is a value which hashes, pickles and copies.

xml 

xml()

Get the record as the element of the annotation.

derivation_of 

derivation_of(model)

Read the record of a model.

Parameters:

Name Type Description Default
model Model

the model.

 required

Returns:

Type Description
Derivation | None

The derivation, None for a model which is not derived.

Raises:

Type Description
ValueError

if the record is not valid.

record_derivation 

record_derivation(model, source, created, targets)

Write the record of a derivation into a model.

A record the model has is extended: the source stays, the created ids and the targets are added. A target which the earlier record created is a created id and not a target.

Parameters:

Name Type Description Default
model Model

the derived model.

 required
source Path

the file of the model the derivation started from.

 required
created Iterable[str]

ids of the parameters which were added.

 required
targets Mapping[str, bool]

id of a parameter which got a rule -> whether it was constant.

 required

Returns:

Type Description
Derivation

The record which was written.

Raises:

Type Description
ValueError

if the record cannot be written.

strip_derivation 

strip_derivation(sbml_path)

Undo the derivation of a model.

Parameters:

Name Type Description Default
sbml_path Path

the derived model.

 required

Returns:

Type Description
SBMLDocument

The document of the source model, i.e. the model without the
Derivation

parameters and rules which were added and without the record, and
tuple[SBMLDocument, Derivation]

the record.

Raises:

Type Description
ValueError

if the model cannot be read, is not derived, a created parameter or a target is not in it, the rule of a target is not the one of the network, or the model refers to a created parameter outside of the parts the derivation wrote, i.e. it was changed by hand after the derivation.