CellMLToolkit.jl is a Julia library that connects CellML models to SciML, the Scientific Julia ecosystem. CellMLToolkit.jl acts as a bridge between CellML and ModelingToolkit.jl. It imports a CellML model (in XML) and emits a ModelingToolkit.jl intermediate representation (IR), which can then enter the SciML ecosystem.
For information on using the package, see the stable documentation. Use the in-development documentation for the version of the documentation, which contains the unreleased features.
CellML is an XML-based open-standard for the exchange of mathematical models. CellML originally started in 1998 by the Auckland Bioengineering Institute at the University of Auckland and affiliated research groups. Since then, its repository has grown to more than a thousand models. While CellML is not domain-specific, its focus has been on biomedical models. Currently, the active categories in the repository are Calcium Dynamics, Cardiovascular Circulation, Cell Cycle, Cell Migration, Circadian Rhythms, Electrophysiology, Endocrine, Excitation-Contraction Coupling, Gene Regulation, Hepatology, Immunology, Ion Transport, Mechanical Constitutive Laws, Metabolism, Myofilament Mechanics, Neurobiology, pH Regulation, PKPD, Protein Modules, Signal Transduction, and Synthetic Biology. There are many software tools to import, process and run CellML models; however, these tools are not Julia-specific.
SciML is a collection of Julia libraries for open source scientific computing and machine learning. The centerpiece of SciML is DifferentialEquations.jl, which provides a rich set of ordinary differential equations (ODE) solvers. One major peripheral component of SciML is ModelingToolkit.jl. It is a modeling framework for high-performance symbolic-numeric computation in scientific computing and scientific machine learning. The core of ModelingToolkit.jl is an IR language to code the scientific problems of interest at a high level. Automatic code generation and differentiation allow for the generation of a usable model for the other components of SciML, such as DifferentialEquations.jl.
To install CellMLToolkit.jl, use the Julia package manager:
using Pkg
Pkg.add("CellMLToolkit")
using CellMLToolkit, DifferentialEquations, Plots
ml = CellModel("models/lorenz.cellml.xml")
prob = ODEProblem(ml, (0,100.0))
sol = solve(prob)
plot(sol, idxs=(1,3)) # idxs keyword has superseded vars keyword
Note that model
is a directory of the CellMLToolkit package. You can find its path as
model_root = joinpath(splitdir(pathof(CellMLToolkit))[1], "..", "models")
and then
model_path = joinpath(model_root, "lorenz.cellml.xml")
ml = CellModel(model_path)
The models directory contains a few CellML model examples. Let's start with a simple one, the famous Lorenz equations.
using CellMLToolkit
ml = CellModel("models/lorenz.cellml.xml")
Now, ml
is a CellModel
structure that contains both a list of the loaded XML files and their components (accessible as ml.doc
) and a ModelingToolkit ODESystem
that defines variables and dynamics and can be accessed as getsys(ml)
.
The next step is to convert ml
into an ODEProblem
, ready to be solved.
prob = ODEProblem(ml, (0,100.0))
Here, (0,100.0)
is the tspan
parameter, describing the integration range of the independent variable.
In addition to the model equations, the initial conditions and parameters are also read from the XML file(s) and are available as prob.u0
and prob.ps
, respectively. We can solve and visualize prob
as
using DifferentialEquations, Plots
sol = solve(prob)
plot(sol, idxs=(1,3)) # idxs keyword has superseded vars keyword
As expected,
Let's look at more complicated examples. The next one is the ten Tusscher-Noble-Noble-Panfilov human left ventricular action potential model. This is a mid-range electrophysiology model with 17 states variables and relatively good numerical stability.
ml = CellModel("models/tentusscher_noble_noble_panfilov_2004_a.cellml.xml")
prob = ODEProblem(ml, (0, 10000.0))
sol = solve(prob, dtmax=1.0) # we need to set dtmax to allow for on-time stimulation
plot(sol, idxs=12)
We can tell which variable to plot (idxs=12
here) by looking at the output of list_states(ml)
(see below).
Let's see how we can modify the initial values and parameters. We will use the Beeler-Reuter model with 8 state variables as an example:
ml = CellModel("models/beeler_reuter_1977.cellml.xml")
The model parameters are listed as list_params(ml)
:
sodium_current₊g_Na => 0.04
sodium_current₊E_Na => 50.0
sodium_current₊g_Nac => 3.0e-5
stimulus_protocol₊IstimStart => 10.0
stimulus_protocol₊IstimEnd => 50000.0
stimulus_protocol₊IstimAmplitude => 0.5
stimulus_protocol₊IstimPeriod => 1000.0
stimulus_protocol₊IstimPulseDuration => 1.0
slow_inward_current₊g_s => 0.0009
membrane₊C => 0.01
Note the form of the parameter and variable names. They are composed of two names joined by a ₊ sign. The first name is the component name in CellML models (e.g., sodium_current, stimulus_protocol) and the second name is the actual variable name.
Similarly, we can list the state variables by calling list_states(ml)
:
slow_inward_current_d_gate₊d(time) => 0.003
slow_inward_current_f_gate₊f(time) => 0.994
slow_inward_current₊Cai(time) => 0.0001
time_dependent_outward_current_x1_gate₊x1(time) => 0.0001
sodium_current_m_gate₊m(time) => 0.011
sodium_current_h_gate₊h(time) => 0.988
membrane₊V(time) => -84.624
sodium_current_j_gate₊j(time) => 0.975
Assume we want to change IstimPeriod
. We can easily do this with the help of update_list!
utility function provided:
params = list_params(ml)
update_list!(params, :stimulus_protocol₊IstimPeriod, 250.0)
prob = ODEProblem(ml, (0, 10000.0); p=last.(params)) # note that you need to pass last.(params) and not params itself to ODEProblem
The rest is the same as before.
sol = solve(prob, dtmax=1.0)
plot(sol, idxs=7) # 7 is the index of membrane₊V
For the next example, we chose a complex model to stress the ODE solvers: the O'Hara-Rudy left ventricular model. This model has 49 state variables, is very stiff, and is prone to oscillation. In the previous versions of this document, we used CVODE_BDF
from the Sundial suite (using Sundials
) to solve this problem. Fortunately, DifferentialEquations.jl has advanced signigficantly such that an efficient and pure Julia solution to the O'Hara-Rudy model is possible.
ml = CellModel("models/ohara_rudy_cipa_v1_2017.cellml.xml")
tspan = (0, 5000.0)
prob = ODEProblem(ml, tspan);
sol = solve(prob, dtmax=1.0)
plot(sol, idxs=49) # membrane₊v
CellML specification allows for models spanning multiple XML files. In these models, the top level CellML XML file imports components from other CellML files, which in turn may import from other files. CellMLToolkit supports this functionality. It assumes that the top-level file and all the imported files reside in the same directory. models/noble_1962
contained one such example:
ml = CellModel("models/noble_1962/Noble_1962.cellml")
prob = ODEProblem(ml, tspan)
sol = solve(prob, dtmax = 0.5)
Note that the syntax is exactly the same as before. However, the list of the imported files are printed during CellModel
generation:
[ Info: importing Noble62_Na_channel.cellml
[ Info: importing Noble62_units.cellml
[ Info: importing Noble62_K_channel.cellml
[ Info: importing Noble62_units.cellml
[ Info: importing Noble62_L_channel.cellml
[ Info: importing Noble62_units.cellml
[ Info: importing Noble62_units.cellml
[ Info: importing Noble62_parameters.cellml
[ Info: importing Noble62_units.cellml
Same as before, we can plot the output as
plot(sol, idxs = 2)