Validating NeuroML Models

Validate NeuroML 2 files before using them.

It is good practice to validate NeuroML 2 files to check them for correctness before using them.

Models described in NeuroML must adhere to the NeuroML specification. This allows all NeuroML models to be checked for correctness: validation. There are a number of ways of validating NeuroML model files.

Using the command line tools

Both pynml (provided by pyNeuroML) and jnml (provided by jNeuroML) can validate individual NeuroML files:

Usage:

# For NeuroML 2
jnml -validate <NML file(s)>
pynml <NML file(s)> -validate

# For NeuroML 1 (deprecated)
jnml -validatev1 <NML file>
pynml <NML file(s)> -validatev1

Using the Python API

The pyNeuroML Python API provides a number of methods to validate NeuroML 2 files. The first is the aptly named validate_neuroml2 function:

from pyneuroml.pynml import validate_neuroml2

...

validate_neuroml2(nml_filename)

Similarly, the validate_neuroml1 function can be used to validate NeuroML v1 files.

If you are loading NeuroML files into your Python script, the read_neuroml2_file function also includes validation:

from pyneuroml.pynml import read_neuroml2_file


....


read_neuroml2_file(nml_filename, include_includes=True, check_validity_pre_include=True)

This will read (load) the provided NeuroML 2 file and all the files that are recursively included by it, and validate them all while it loads them.

List of validation tests

These tests are made against the Schema document.

Test	Description
Check names	Check that names of all elements, attributes, parameters match those provided in the schema
Check types	Check that the types of all included elements
Check values	Check that values follow given restrictions
Check inclusion	Check that required elements are included
Check cardinality	Check the number of elements
Check hierarchy	Check that child/children elements are included in the correct parent elements
Check sequence order	Check that child/children elements are included in the correct order

These are additional validation tests that are run on models (defined here):

Test	Description
Check top level ids	Check that top level (root) elements have unique ids
Check Network level ids	Check that child/children of the Network element have unique ids
Check Cell Segment ids	Check that all Segments in a Cell have unique ids
Check single Segment without parent	Check that only one Segment is without parents (the soma Segment)
Check SegmentGroup ids	Check that all SegmentGroups in a Cell have unique ids
Check Member segment ids exist	Check that Segments referred to in SegmentGroup Members exist
Check SegmentGroup definition	Check that SegmentGroups being referenced are defined
Check SegmentGroup definition order	Check that SegmentGroups are defined before being referenced
Check included SegmentGroups	Check that SegmentGroups referenced by Include elements of other SegmentGroups exist
Check numberInternalDivisions	Check that SegmentGroups define numberInternalDivisions (used by simulators to discretize un-branched branches into compartments for simulation)
Check included model files	Check that model files included by other files exist
Check Population component	Check that a component id provided to a Population exists
Check ion channel exists	Check that an ion channel used to define a ChannelDensity element exists
Check concentration model species	Check that the species used in ConcentrationModel elements are defined
Check Population size	Check that the size attribute of a PopulationList matches the number of defined Instances
Check Projection component	Check that Populations used in the Projection elements exist
Check Connection Segment	Check that the Segment used in Connection elements exist
Check Connection pre/post cells	Check that the pre- and post-synaptic cells used in Connection elements exist and are correctly specified
Check Synapse	Check that the Synapse component used in a Projection element exists
Check root id	Check that the root Segment in a Cell morphology has id (0)