Hierarchical: restructuring and add relative to InnerCalculatorCollector #1245
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Inner parameters were not being stored when the result was stored. This was due to a problem with putting a dictionary into an HDF5 format. I've transformed it into 2 lists now: a INNER_PARAMETERS_VALUES and INNER_PARAMETER_NAMES list.
- Restructuring of the hierarchical scaling+offset method. The base classes used by all non-quantitative and semi-quantitative data types are put in the `base_...` files. All classes related to the scaling+offset method (for relative data) is moved to the `relative` folder, analogous to other data types. - Renaming: the naming scheme doesn't follow the method, but the data type now. It's more direct and removes the ambiguity of `OptimalScaling` for ordinal data and `scaling_offset` for relative data. - Optimal scaling -> ordinal - Spline approximation -> semiquantitative - Scaling + offset -> relative
Codecov ReportAttention:
❗ Your organization needs to install the Codecov GitHub app to enable full functionality. Additional details and impacted files@@ Coverage Diff @@
## develop #1245 +/- ##
===========================================
- Coverage 84.59% 84.28% -0.32%
===========================================
Files 148 151 +3
Lines 12122 12331 +209
===========================================
+ Hits 10255 10393 +138
- Misses 1867 1938 +71 ☔ View full report in Codecov by Sentry. |
Doresic
requested review from
dweindl,
stephanmg,
m-philipps,
FFroehlich,
dilpath and
a team
as code owners
FFroehlich
approved these changes
Jan 8, 2024
doc/example/example_semiquantitative/measurementData_example_semiquantitative_linear.tsv
Show resolved
Hide resolved
dweindl
approved these changes
Jan 8, 2024
stephanmg
approved these changes
Jan 9, 2024
|
Is there an option to pre-view the documentation? |
Yes, it should be possible by clicking on the readthedocs test and then view Docs (if it runs successfully). @m-philipps |
dweindl
reviewed
Jan 9, 2024
dilpath
reviewed
Jan 10, 2024
dilpath
approved these changes
Jan 10, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Complete restructuring of the hierarchical module.
The formerly named
hierarchicalmodule which optimizes scaling, offset, and noise of relative data was renamed torelative. Classes used by all non-quantitative data type hierarchical algorithms (ordinal, relative, and semiquantitative) were moved tobase_...files, while the classes and functions related specifically to relative data (scaling and offset problem, calculator, solver etc...) were moved to the hierarchicalrelativesub-module.Renamed the former
optimal_scalingmodule used for ordinal data toordinalto remove any confusion regarding a similar name of scaling parameters for relative data.Renamed the former
spline_approximationmodule tosemiquantitative. Now all hierarchical submodules are called after the data they integrate with rather than the method they use.A lot of Class renaming to fit the renaming already mentioned.
Included the
relativemethod (scaling and offset) into the InnerCalculatorCollector. Now, one can have a PEtab problem containing all three types of observables: relative, ordinal, and semiquantitative observables.Added hierarchical PEtab validation to prevent things like specifying relative observables with semiquantitative data etc...
Fixed visualizations, docs, and tests to work with the renaming.
The
RelativeCalculatorhas two operating modes:call_amici_twiceandcalculate_directly. The former is the old implementation of the calculator. It simulates the model twice. However, this can be used only if relative data is the only non-quantitative data type. In this case, one can use 2nd order sensitivities and adjoint sensitivity analysis as before. These two are not implemented for the latter modecalculate_directly. This mode will simulate only once, and calculate the objective function and gradient directly (but only for the relative data, not other data). It can be used only with forward sensitivity analysis since it uses sensitivities in the gradient calculation.