Merge pull request #769 from SciML/structuralidentifiabilityextension

Structuralidentifiabilityextension

HISTORY.md

@@ -3,6 +3,19 @@
 ## Catalyst unreleased (master branch)
 
 ## Catalyst 14.0
+- Added CatalystStructuralIdentifiabilityExtension, which permits StructuralIdentifiability.jl function to be applied directly to Catalyst systems. E.g. use
+```julia
+using Catalyst, StructuralIdentifiability
+goodwind_oscillator = @reaction_network begin
+    (mmr(P,pₘ,1), dₘ), 0 <--> M
+    (pₑ*M,dₑ), 0 <--> E
+    (pₚ*E,dₚ), 0 <--> P
+end
+assess_identifiability(goodwind_oscillator; measured_quantities=[:M])
+```
+to assess (global) structural identifiability for all parameters and variables of the `goodwind_oscillator` model (under the presumption that we can measure `M` only).
+- Automatically handles conservation laws for structural identifiability problems (eliminates these internally to speed up computations).
+- Adds a tutorial to illustrate the use of the extension.
 - Simulation of spatial ODEs now supported. For full details, please see https://github.com/SciML/Catalyst.jl/pull/644 and upcoming documentation. Note that these methods are currently considered alpha, with the interface and approach changing even in non-breaking Catalyst releases.
 - LatticeReactionSystem structure represents a spatial reaction network:
   ```julia

Project.toml

@@ -26,10 +26,12 @@ Unitful = "1986cc42-f94f-5a68-af5c-568840ba703d"
 [weakdeps]
 BifurcationKit = "0f109fa4-8a5d-4b75-95aa-f515264e7665"
 HomotopyContinuation = "f213a82b-91d6-5c5d-acf7-10f1c761b327"
+StructuralIdentifiability = "220ca800-aa68-49bb-acd8-6037fa93a544"
 
 [extensions]
 CatalystBifurcationKitExtension = "BifurcationKit"
 CatalystHomotopyContinuationExtension = "HomotopyContinuation"
+CatalystStructuralIdentifiabilityExtension = "StructuralIdentifiability"
 
 [compat]
 BifurcationKit = "0.3"
@@ -48,6 +50,7 @@ Reexport = "0.2, 1.0"
 Requires = "1.0"
 RuntimeGeneratedFunctions = "0.5.12"
 Setfield = "1"
+StructuralIdentifiability = "0.5.1"
 SymbolicUtils = "1.0.3"
 Symbolics = "5.14"
 Unitful = "1.12.4"
@@ -69,8 +72,9 @@ StableRNGs = "860ef19b-820b-49d6-a774-d7a799459cd3"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 SteadyStateDiffEq = "9672c7b4-1e72-59bd-8a11-6ac3964bc41f"
 StochasticDiffEq = "789caeaf-c7a9-5a7d-9973-96adeb23e2a0"
+StructuralIdentifiability = "220ca800-aa68-49bb-acd8-6037fa93a544"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 Unitful = "1986cc42-f94f-5a68-af5c-568840ba703d"
 
 [targets]
-test = ["BifurcationKit", "DomainSets", "Graphviz_jll", "HomotopyContinuation", "NonlinearSolve", "OrdinaryDiffEq", "Random", "SafeTestsets", "SciMLBase", "SciMLNLSolve", "StableRNGs", "Statistics", "SteadyStateDiffEq", "StochasticDiffEq", "Test", "Unitful"]
+test = ["BifurcationKit", "DomainSets", "Graphviz_jll", "HomotopyContinuation", "NonlinearSolve", "OrdinaryDiffEq", "Random", "SafeTestsets", "SciMLBase", "SciMLNLSolve", "StableRNGs", "Statistics", "SteadyStateDiffEq", "StochasticDiffEq", "StructuralIdentifiability", "Test", "Unitful"]

docs/Project.toml

@@ -7,6 +7,7 @@ Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
 Documenter = "e30172f5-a6a5-5a46-863b-614d45cd2de4"
 HomotopyContinuation = "f213a82b-91d6-5c5d-acf7-10f1c761b327"
 Latexify = "23fbe1c1-3f47-55db-b15f-69d7ec21a316"
+Logging = "56ddb016-857b-54e1-b83d-db4d58db5568"
 ModelingToolkit = "961ee093-0014-501f-94e3-6117800e7a78"
 NonlinearSolve = "8913a72c-1f9b-4ce2-8d82-65094dcecaec"
 Optim = "429524aa-4258-5aef-a3af-852621145aeb"
@@ -22,6 +23,7 @@ Setfield = "efcf1570-3423-57d1-acb7-fd33fddbac46"
 SpecialFunctions = "276daf66-3868-5448-9aa4-cd146d93841b"
 SteadyStateDiffEq = "9672c7b4-1e72-59bd-8a11-6ac3964bc41f"
 StochasticDiffEq = "789caeaf-c7a9-5a7d-9973-96adeb23e2a0"
+StructuralIdentifiability = "220ca800-aa68-49bb-acd8-6037fa93a544"
 Symbolics = "0c5d862f-8b57-4792-8d23-62f2024744c7"
 
 [compat]
@@ -47,4 +49,5 @@ Setfield = "1.1"
 SpecialFunctions = "2.1"
 SteadyStateDiffEq = "2.0.1"
 StochasticDiffEq = "6"
+StructuralIdentifiability = "0.5.1"
 Symbolics = "5.14"

docs/pages.jl

@@ -16,6 +16,7 @@ pages = Any["Home" => "index.md",
                                            "catalyst_applications/nonlinear_solve.md",
                                            "catalyst_applications/bifurcation_diagrams.md"],
             "Inverse Problems" => Any["inverse_problems/parameter_estimation.md",
-                                      "inverse_problems/petab_ode_param_fitting.md"],
+                                      "inverse_problems/petab_ode_param_fitting.md",
+                                      "inverse_problems/structural_identifiability.md"],
             "FAQs" => "faqs.md",
             "API" => "api/catalyst_api.md"]

docs/src/inverse_problems/petab_ode_param_fitting.md

@@ -127,7 +127,7 @@ fitted_sol = solve(oprob_fitted, Tsit5())
 plot!(fitted_sol; idxs=:P, label="Fitted solution", linestyle=:dash, lw=6, color=:lightblue)
 ```
 
-Here we use the `get_ps` function to retrieve a full parameter set using the optimal parameters. Alternatively, the `ODEProblem` or fitted simulation can be retrieved directly using the `get_odeproblem` or `get_odesol` [functions](https://sebapersson.github.io/PEtab.jl/dev/API_choosen/#PEtab.get_odeproblem), respectively (and the initial condition using the `get_u0` function). The calibration result can also be found in `res.xmin`, however, note that PEtab automatically ([unless a linear scale is selected](@ref petab_parameters_scales)) converts parameters to logarithmic scale, so typically `10 .^res.xmin` are the values of interest. If you investigate the result from this example you might note, that even if PEtab.jl has found the global optimum (which fits the data well), this does not actually correspond to the true parameter set. This phenomenon is related to the concept of *identifiability*, which is very important for parameter fitting.
+Here we use the `get_ps` function to retrieve a full parameter set using the optimal parameters. Alternatively, the `ODEProblem` or fitted simulation can be retrieved directly using the `get_odeproblem` or `get_odesol` [functions](https://sebapersson.github.io/PEtab.jl/dev/API_choosen/#PEtab.get_odeproblem), respectively (and the initial condition using the `get_u0` function). The calibration result can also be found in `res.xmin`, however, note that PEtab automatically ([unless a linear scale is selected](@ref petab_parameters_scales)) converts parameters to logarithmic scale, so typically `10 .^res.xmin` are the values of interest. If you investigate the result from this example you might note, that even if PEtab.jl has found the global optimum (which fits the data well), this does not actually correspond to the true parameter set. This phenomenon is related to the [concept of *identifiability*](@ref structural_identifiability), which is very important for parameter fitting.
 
 ### Final notes
 PEtab.jl also supports [multistart optimisation](@ref petab_multistart_optimisation), [automatic pre-equilibration before simulations](https://sebapersson.github.io/PEtab.jl/stable/Brannmark/), and [events](@ref petab_events). Various [plot recipes](@ref petab_plotting) exist for investigating the optimisation process. Please read the [PETab.jl documentation](https://sebapersson.github.io/PEtab.jl/stable/) for a more complete description of the package's features. Below follows additional details of various options and features (generally, PEtab is able to find good default values for most options that are not specified).