Biological Models

Biological Models

The biological models functionality is provided by DiffEqBiological.jl and helps the user build discrete stochastic and differential equation based systems biological models. These tools allow one to define the models at a high level by specifying reactions and rate constants, and the creation of the actual problems is then handled by the modeling package.

The Reaction Type

The basic type for BiologicalModels.jl is the reaction type. Its constructor is:


rate_constant is the rate constant for the reaction. reactants is a list of reactants for the reaction. For example, reactants=(2,3) means that the reaction rate is rate_constant*u[2]*u[3]. stoichiometry is then the stoichiometry for the result. It is a list of tuples of changes to apply when the reaction takes place. Each tuple (i,j) means "modify reactiant i by amount j". For example, the tuple (2,-1) means "decrease reactant 2 by 1".

Note About Rate Dependency

Note that currently, the reactions are used to build ConstantRateJumps. This means that the solver requires that the rates are constant between jumps in order to achieve full accuracy. The rates for the ConstantRateJump may depend on each other, but they may not depend on the differential equation themselves.

Variable Rate Reactions

VariableRateReaction are allowed to have their rates change continuously, depending on time or values related to a differential equation. The constructor is:

function VariableRateReaction(rate_constant,reactants,stoichiometry;
                              idxs = nothing,

The additional keyword arguments are for controlling the associated ContinuousCallback used to handle VariableRateReactions in simulations.

Example Reaction

An example reaction is:

r1 = Reaction(1e-4,(1,2),((1,-1),(2,1)))

Here, the rate_constant is 1e-4. The reactants are components 1 and 2, meaning the reaction rate is calculated by rate_constant*u[1]*u[2]. The stoichiometry does two things. First, the (1,-1) means that, when the reaction occurs, we decrease u[1] by 1. Secondly, the (2,1) means we increase u[2] by 1. Thus this reaction is a reaction where chemical 1 changes into chemical 2, and it is enhanced by chemical 2 itself.


These reactions can be added to a differential equation (or discrete) problem using the GillespieProblem. This is simply a constructor which interprets the reactions as jumps, and builds the associated JumpProblem. Thus its constructor is the same:


This is the exact same constructor as the JumpProblem, except now we pass reactions (or VariableRateReactions, or a ReactionSet) instead of jumps. Thus for more information, see the description of the JumpProblem.