Model format¶

The model should be written in a plain text file, and should include all of the headings in the block below. Note that all # are required, but it is not necessary that there exists text in each heading only that the heading is present.

########## NAME
########## METADATA
########## MACROS
########## STATES
########## PARAMETERS
########## VARIABLES
########## FUNCTIONS
########## EVENTS
########## OUTPUTS
########## INPUTS
########## FEATURES

What should go under the different headings?¶

Below, we will detail what information should go in the different headings.

NAME¶

Under the Name heading, the model name should be defined. Assume that the model is called my_model

########## NAME
my_model

Then, after installation, the model will be importable by:

sund.import_model('my_model')

METADATA¶

Under the METADATA heading, metadata regarding the model can be entered. The only mandatory information is the time unit of the model. For a list of possible time units, see reference/Time units. Below is an example for a model defined in seconds:

########## METADATA
time_unit = s

STATES¶

Here, we define the ODEs and the initial conditions for each state. The ODEs can be defined in two different ways, either d/dt(X) or ddt_X. The initial conditions are defined as X(0)=1 (the initial value of X is 1).

Below is an example of two states A and B which have two reaction where A → B and B → A:

########## STATES
ddt_A = k1*B - k2*A
d/dt(B) = - k1*B + k2*A

A(0) = 10
B(0) = 20

PARAMETERS¶

Here, values for the model parameters should be listed. All parameters used in the ODEs should be listed.

Below is an example for the ODEs defined above:

########## PARAMETERS
k1 = 2
k2 = 3

VARIABLES¶

Here, any mathematical expression can be defined. This could be for example to calculate som property based on the parameter values, or to define the reaction rates as a variable.

Below is an example, where the rates for A and B is expressed as variables instead.

########## STATES
ddt_A =  v1 - v2
d/dt(B) = - v1 + v2
....
########## VARIABLES
v1 = k1*B
v2 = k2*A

Events¶

An event corresponds to a change that happen at a given condition. The events follow the formalism with: event_name = condition, propert1, value1, property2, value2.

Below is an example where the insulin production (ins_prod) start while the plasma glucose (glu) is above 5, and stops when it is below 5:

########## EVENTS
event_start_insulin = glu>=5, ins_prod, 1
event_stop_insulin = glu<5, ins_prod, 0

It is possible to do time-based events by using the in-built variable time ( e.g. time>15). But, if the time when an event happens is already known, it is better to use a predefined activity instead.

INPUTS¶

Defines the input that the model needs from either another model, or an activity object. A default value can be defined to be used if the input is not supplied. The default value is defined using @ value.

Below is an example where the model needs an input epi which corresponds to the property (state, parameter, variable etc.) epinephrine_extracellular from another model. If no epinephrine_extracellular, assume epi to be zero.

########## INPUTS
epi = epinephrine_extracellular @ 0

OUTPUTS¶

Outputs are properties (state, parameter, variable etc.) that can be shared with other models. These can have different internal names compared to the external name shown to other models.

Assume that we have a model which provides the epinephrine_extracellular to other models, and that the internal name for the property is adrenaline:

########## OUTPUTS
epinephrine_extracellular = adrenaline

It is also possible to provide multiple outputs from the same internal property, useful when e.g. using different nomenclatures:

########## OUTPUTS
epinephrine_extracellular = adrenaline
adrenaline_extracellular = adrenaline

FEATURES¶

Lists whatever can be observed in the model. Can include name changes, to convert from internal names of e.g. a steady to something readable. The feature names and feature data of the defined features will show up in the simulation.feature_names and simulation.feature_data after a simulation has been done. A feature can correspond to any model property (state, parameter, variable etc.).

Assume that we want to have two different features insulin and epinephrine which corresponds to the internal states ins and epi:

########## OUTPUTS
insulin = insulin
epinephrine = epinephrine