Add terms for (in)dependent variables of outputs of simulations

[jonrkarr]

@fbergmann @matthiaskoenig I'm building up a list of necessary terms here

• kinetic (e.g., SBML)
  • [x] time (urn:sedml:symbol:time)
  • [x] concentration
  • [x] amount
  • [x] particle number
  • [x] rate
  • [x] particle number rate
• flux balance (e.g., SBML-fbc)
  • [x] (objective) value
  • [x] flux
  • [x] reduced costs
  • [x] shadow prices
  • [x] min flux
  • [x] max flux
• logical (e.g., SBML-qual)
  • [x] step?
  • [x] level

[jonrkarr]

Here's the proposed organization:

[fbergmann]

that looks fine to me. (I might have chosen the names lower / upper bound for the flux for fbc problems, but it is fine to me).

[jonrkarr]

I can add those as synonyms -- another benefit of organizing this information as an ontology.

On Fri, Mar 26, 2021, 6:18 AM Frank Bergmann @.***> wrote:

that looks fine to me. (I might have chosen the names lower / upper bound for the flux for fbc problems, but it is fine to me).

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://github.com/SED-ML/KiSAO/issues/85#issuecomment-808097472, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAVXMKLXTAB2DI56CD4H4JTTFRNPBANCNFSM4ZZZNSTA .

[hsauro]

Within the Jacobian matrix, there should also be reduced and full Jacobian matrix. Also in the sensitivities we need:

Concentration Control Coefficients (scaled and unscaled) Flux Control Coefficients (scaled and unscaled) Response Coefficient (scaled and unscaled) Elasticities (scaled and unscaled)

Also don't forget eigenvalues and eigenvectors.

To make it complete we should also add (possibly under a different heading such as 'Structural' or similar.

Stoichiometry matix Reduced stoichiometry matrix Link matrix Conservation Matrix

Herbert

On Thu, Mar 25, 2021 at 12:19 PM Jonathan Karr @.***> wrote:

Here's the proposed organization: [image: Screenshot from 2021-03-25 15-16-53] https://user-images.githubusercontent.com/2848297/112530894-53b91900-8d7d-11eb-8e5b-71e227b0c02c.png

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/SED-ML/KiSAO/issues/85#issuecomment-807319065, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAIBSDWFYW53ASBSKURTM7TTFOECVANCNFSM4ZZZNSTA .

-- Herbert Sauro, Professor University of Washington, Bioengineering 206-685-2119, www.sys-bio.org @.*** Books: http://books.analogmachine.org/

[matthiaskoenig]

It would be great if we could have amount and concentration on the same level. A concentration is not really an amount, so it would be great to have a more general term. E.g. extent so the tree would look like this

• extent
  • amount
  • concentration
  • level
  • particle number

What is propensity in this context?

The variable dimension slice seems not to fit in there. I think we need some more discussion on this part before deciding this. E.g. what is the difference between start/initial and end/final. I am not sure the slicing operations belong in the ontology.

[jonrkarr]

Extent/amount is easy to change.

Happy remove slicing and merge this into the variable branch with a combinatorial number of terms. Let's discuss slicing at the next SED-ML meeting to make sure we all align on 1 design.

[luciansmith]

I feel like the two things you would need are 'current' and 'initial' (i.e. not 'end', per se.) However, if you queried the 'current' condition of a variable before simulation, that would give you the 'initial', too, so potentially would we only need 'current'? It could easily be that I'm not envisioning the same scenarios that the rest of you are.

-Lucian

On Fri, Mar 26, 2021 at 3:17 PM Jonathan Karr @.***> wrote:

Extent/amount is easy to change.

Happy remove slicing and merge this into the variable branch with a combinatorial number of terms. Let's discuss slicing at the next SED-ML meeting to make sure we all align on 1 design.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/SED-ML/KiSAO/issues/85#issuecomment-808556693, or unsubscribe https://github.com/notifications/unsubscribe-auth/AANH3VXRDFFAFANSRUAUJHLTFUBVHANCNFSM4ZZZNSTA .

[maxneal]

I wanted to chime in here because it looks like many of the terms that are slated to be added to KiSAO in this thread are already represented and organized in the Ontology of Physics for Biology (OPB). For example, the OPB represents the concepts of time, concentration, particle number, and concentration/particle rates. In the interest of integrating knowledge resources across the COMBINE community and not duplicating content across these resources, it would probably be a good idea to ensure that KiSAO and the OPB are appropriately integrated. For example, OPB terms could be used directly within the KiSAO hierarchy where appropriate, or KiSAO terms could be mapped to their related OPB terms. Additionally, you could leverage the OPB's hierarchical organization to address issues such as the one Matthias raised in his most recent comment about classifying amount and concentration (see OPB:Amount property and its descendants). The OPB is intended to serve the community as a core ontology that represents - among other things - the physical properties simulated in biosimulation models (for example, it is the recommended ontology for representing physical properties in the OMEX metadata specification), and many of the terms slated for addition to KiSAO here are in the physical property tree of the OPB. We are happy to collaborate further on this to ensure useful integration between KiSAO and the OPB. @jhgennari @dcookUW

[jonrkarr]

This is one of the reasons I pushed to organize this information into an ontology -- to link all of the terms introduced into KiSAO to other ontologies, in particular OPB and SIO. In contrast, SED-ML's current approach of URNs doesn't provide a clear path to achieving something similar.

I thought about using OPB, SIO, or another ontology directly in SED-ML documents. In the context of SED-ML, these terms would be used to define the computational execution of SED-ML documents. To use OPB or other terms directly in SED-ML documents, their mathematical meaning would need to be captured inside the ontology and there would need to be a standard way for simulation tools to translate each term into computational operations. For example, the OPB term for concentration would need to define the meaning of concentration as a mathematical expression of amount divided by volume and Avogadro's number. As I understand, the OPB doesn't capture this.

Instead, I thought a good way to manage this is to create a relatively small list of terms in KiSAO for which is there a significant degree of overlapping support among simulation tools (that can execute similar models). Essentially, this list of terms could be a subset of those in OPB and/or SIO, where KiSAO is used to track the supported subset for which simulation tools have a shared understanding of their computational meaning. Embracing all of the terms in OPB seems hard because this would likely lead to SED-ML documents that can only be interpreted by specific tools that recognize specific terms.

Adding mathematical meaning to OPB would be very useful. This would help solve this problem. I would be interested to discuss how to apply this to SED-ML.

[maxneal]

Jonathan - could you provide a use case example that illustrates how attaching a mathematical meaning to an OPB term would be useful? I'm not super familiar with the end goals here nor how classifying model variables as concentrations, etc. would be used to support the computational execution of SED-ML docs. I'm trying to figure out whether it would make sense to add mathematical definitions to the OPB to support these efforts. Or maybe it would make more sense to include those definitions on the KiSAO terms and just link them to OPB terms via equivalent class or subclass-of statements.

[jonrkarr]

Moved to dependent-variables branch