Background
This issue summarises the question of whether the mapping table should be allowed to contribute additional valid entries to the parameter table. This came up during review of #482 and #480.
For currently supported model types (SBML, PySB), the assumption has been that the mapping table would not introduce additional valid parameter table entries.
However, Petab-SciML requires this kind of mapping table contribution. For example, a network parameter group net1.parameters could be assigned a PEtab ID (e.g. net_ps) in the mapping table, and that ID would then appear in the parameters table with its bounds and priors. In this case, the mapping table does need to feed into parameter table validation.
Open questions
- Are SciML problems the only use case where mapping table entries need to be treated as valid parameter table parameters?
- What API would be needed to validate parameters contributed by the mapping table?
- A SciML specific
ValidationTask is one option
Next steps
- Confirm whether SciML is the only use case for this requirement
- Design the API to support it
- Implement and test
Background
This issue summarises the question of whether the mapping table should be allowed to contribute additional valid entries to the parameter table. This came up during review of #482 and #480.
For currently supported model types (SBML, PySB), the assumption has been that the mapping table would not introduce additional valid parameter table entries.
However, Petab-SciML requires this kind of mapping table contribution. For example, a network parameter group
net1.parameterscould be assigned a PEtab ID (e.g.net_ps) in the mapping table, and that ID would then appear in the parameters table with its bounds and priors. In this case, the mapping table does need to feed into parameter table validation.Open questions
ValidationTaskis one optionNext steps