Tellurium l1v4 compliant! Ish!

[Lucian Smith]

I now have a version of Tellurium that implements support for the majority of new SED-ML L1v4 features.  We've discussed the plotting extensively before; I now have support for a large number of examples of various types of analyses, including several Metabolic Control Analysis functions and Stoichiometry analysis functions.

All of the SED-ML files I created (along with the python scripts I used to make them) can be found in 

https://github.com/SED-ML/sedml-test-suite/tree/master/contributions/HARMONY%202021

and one test in particular that does a bunch of MCA analysis is in:

https://github.com/SED-ML/sedml-test-suite/tree/master/contributions/HARMONY%202021/metabolic%20control%20analysis

The full list of the URNs I implemented support for is at:

https://docs.google.com/spreadsheets/d/1f3rBUUhOsB6k7-i_ins0nlB7PTtd_huvhgFrCOh501U/edit#gid=0

I encourage anyone with a SED-ML interpreter to look these up and tell me if you'd interpret them differently than I!

Remaining issues:

* As Jonathan pointed out, if you ask for 'the Jacobian' from a timecourse task, one reasonable interpretation is that you want the Jacobian output at every time point.  I didn't do this, and I'm not sure how to convey this to a SED-ML user.  The options that I can think of are:  1) we say whether it's an every-time-point analysis or an at-the-end analysis explicitly per analysis URN on the website, or 2) we add a new attribute that says which one to do.

* I made up a bunch of sedml-based URNs; do we want KiSAO instead?

* There are still a handful of items on https://docs.google.com/spreadsheets/d/1-7uRLppjiGZ1jyhEDRhpWLDi9iR9Oi9cBQjplv0OBnk/edit?ts=605a2d23#gid=0 with no support from any tool; we should start looking into either finding groups that want to support those elements or just drop them.  The biggest gap is that I have no information about any tool supporting parameter estimation.  Frank?

-Lucian

[Jonathan Karr]

Nice progress!

How are the reduction functions (e.g., urn:sedml:function:average) intended to capture the dimension(s) that the reduction should be applied to? E.g., how would one indicate "average of time" vs "average of iterations of a repeated task"?

My preference would be to add these symbols to KiSAO so (a) there's a machine-readable version of this information in a standard format and (b) there's a place to browse this information (e.g., BioPortal, OLS) that we don't have to maintain.

Much of the new L1V4 could be added to our SED-ML test fairly easily -- e.g., new plots, other algorithms, new symbols. Any additions would be applicable to testing all model languages, and then the deployment is already setup. Would you be interested to work together on it?

Jonathan

--
You received this message because you are subscribed to the Google Groups "sed-ml-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sed-ml-discus...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/sed-ml-discuss/CAHLmBr2greomPtFq-HzksQY_rVbiWcJAz%3DhoOut5_TSqORf0cA%40mail.gmail.com.

[Lucian Smith]

On Mon, May 31, 2021 at 2:12 PM Jonathan Karr <jonr...@gmail.com> wrote:

Nice progress!

How are the reduction functions (e.g., urn:sedml:function:average) intended to capture the dimension(s) that the reduction should be applied to? E.g., how would one indicate "average of time" vs "average of iterations of a repeated task"?

That's handled by the RemainingDimension class--see https://github.com/SED-ML/sedml-test-suite/blob/master/contributions/HARMONY%202021/stoch_remaining_dimensions_avg_max_min_std.sedml

My preference would be to add these symbols to KiSAO so (a) there's a machine-readable version of this information in a standard format and (b) there's a place to browse this information (e.g., BioPortal, OLS) that we don't have to maintain.

That would be fine with me.  I do like being able to read the XML itself in a pinch, so I think I would like to propose that we highly recommend setting the 'name' of all elements with KiSAO terms to be the name of the term.  I think I'll add a function in libsedml to do this automatically with 'setKisao' using a lookup table.

Much of the new L1V4 could be added to our SED-ML test fairly easily -- e.g., new plots, other algorithms, new symbols. Any additions would be applicable to testing all model languages, and then the deployment is already setup. Would you be interested to work together on it?

That would be great!  Everything in the contributions/HARMONY folder is fair game to add, though the expected output isn't always well organized.  I'm more than happy to work with you to set that up.

-Lucian

[Jonathan Karr]

We could directly add these examples to the test suite. If we build the BioSimulators tellurium image with a new version of tellurium, we can also display examples there.

To test all combinations of SED-ML + model languages + simulation classes + simulation algorithms, we'd need to organize this as code that generates SED-ML dynamically for the new L1V4 features. These dynamically generated tests are the main part of the test suite. We already have several base classes to build from. This takes more work to set up, but makes it easy to extend the test suite to more model languages and algorithms. I should probably focus on getting BioModels and other things working well with L1V3 before focusing on L1V4. I'd be happy to orient you if you'd like to add cases.

Jonathan

Jonathan

--

You received this message because you are subscribed to the Google Groups "sed-ml-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sed-ml-discus...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/sed-ml-discuss/CAHLmBr2yeURDCfbbVwoxr-hZUqPpWr6_rvOnO8tmdOXdSA-ALg%40mail.gmail.com.