Improve librr api #362
Improve librr api #362
Conversation
- introduce `update_time_step` in `RoadRunnerIntracellular`
- set for each instance of `RoadRunnerIntracellular`
- this controls the timestep for the model
- will be set to `intracellular_dt` if not specified in the xml for the given cell type
- add the `intracellular_dt` element as a child of the `intracellular` element in the xml
- could be set on a cell-by-cell basis if desired
- introduce `previous_update_time` in `RoadRunnerIntracellular`
- tracks the previous update time
- used to determine how long to run the ODE when updating
- set to `current_time` when copying/cloning
- fix `next_librr_run` in `RoadRunnerIntracellular`
- was not being updated and thus ODE models were running at every time step
- check `current_time >= next_librr_run - 0.5 * diffusion_dt` to determine if an update is needed
- cleaned up librr files
- fixed and cleand ode-energy-sample
- it was running the intracellular models in `main.cpp` and in the update cells function
- it seemed to try to set intracellular_dt inside the intracellular element of the xml
- messy custom.cpp and custom.h
a low-level implementation of the delay differential equation solver
before, only new ones were validated. those "copied" from previously read defs were not Presumably, it was assumed that they fully inherited their parent's model, but as of now, they read the xml to get their model, so they need to be validated.
- `<map type="output" physicell_name="custom:intra_energy" sbml_species="Energy"></map>` - `type` is any of `input`, `output`, `io` to indicate whether it is before or after (or both) the update - `physicell_name` can be either a signal/behavior or one of the previous tokens - `sbml_species` is the name of the species in the SBML model - inputs are validated to ensure all sbml_species are unique - outputs are validated to ensure all physicell_names are unique - on `start()`, any signals/behaviors are validated to be in the relevant dictionary
- also output helpful info when using old, redundant tokens (non-volume ones)
multiple cells would be vying to all set the same voxel concentration - also, clean up ode-energy-sample custom.cpp
|
Still reviewing, but really liking this so far! Obviously, we'll need to synchronize changes for the Studio at release. I do wonder if we should expand the scope of this sample project a bit in order to demo/test for other options, e.g., Unless I'm missing something, I don't think we want |
drbergman
force-pushed
the
improve-librr-api
branch
from
7475c27 to
f52f008
Compare
|
Great! And roger that on the studio updates. Hopefully pretty straightforward since we're now using signals/rules. I think that's a good idea to at least have one On |
As of opening this PR, this is ahead of the branch for PR #361 and is a bigger, better overhaul of the libRoadRunner API than that PR. Because it is bigger, I decided to make it a separate PR so interested parties could compare. Where #361 is refactoring/cleaning/fixing, this one would be a major change. It is backwards compatible except for minor modifications to config files setting intracellular
mappings for libRoadRunner. See below.Importantly, this update makes many of the mappings automatic with only modifying the config file, i.e., no custom coding necessary (users would still need to ensure the
Makefilehas the libRoadRunner macro, compile target, and other updates (see thesample_projects_intracellular/ode/ode_energy/Makefilefor example). Look at the updatedcustom.cppfor theode_energyproject and observe how much simpler this is now!Updated xml
A typical previous child element of the
<intracellular>element would look likewhere
<key>could be any ofsubstrate,phenotype, orcustom_data.Now, that element would be
Some notes:
typeis any ofinput,output, orioand defines whether the mapping is applied before or after updating the intracellular model AND whether PhysiCell values are being input into the model or output from the model (or both in the case ofio)physicell_nameintracellular <substrate_name>(assume sbml species is concentration and sets the total internalized using current cell volume)volume(rescales all volumes)damagevtsc,vtsn, andvff) should be considered best practiceoutputorio), thephysicell_name's must be uniquesbml_speciesinputorio), thesbml_species's must be unique