The deterministic and stochastic approaches

Stochastic simulation algorithms

Comparing stochastic simulation and ODEs
Modelling challenges
An Introduction to Stochastic Simulation
Stephen Gilmore
Laboratory for Foundations of Computer Science
School of Informatics
University of Edinburgh
PASTA workshop, London, 29th June 2006
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background
The modelling of chemical reactions using deterministic rate
laws has proven extremely successful in both chemistry and
biochemistry for many years.
This deterministic approach has at its core the law of mass
action, an empirical law giving a simple relation between
reaction rates and molecular component concentrations.
Given knowledge of initial molecular concentrations, the law
of mass action provides a complete picture of the component
concentrations at all future time points.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background
The modelling of chemical reactions using deterministic rate
laws has proven extremely successful in both chemistry and
biochemistry for many years.
This deterministic approach has at its core the law of mass
action, an empirical law giving a simple relation between
reaction rates and molecular component concentrations.
Given knowledge of initial molecular concentrations, the law
of mass action provides a complete picture of the component
concentrations at all future time points.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background
The modelling of chemical reactions using deterministic rate
laws has proven extremely successful in both chemistry and
biochemistry for many years.
This deterministic approach has at its core the law of mass
action, an empirical law giving a simple relation between
reaction rates and molecular component concentrations.
Given knowledge of initial molecular concentrations, the law
of mass action provides a complete picture of the component
concentrations at all future time points.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background: Law of Mass Action
The law of mass action considers chemical reactions to be
macroscopic under convective or diusive stirring, continuous
and deterministic.
These are evidently simplications, as it is well understood
that chemical reactions involve discrete, random collisions
between individual molecules.
As we consider smaller and smaller systems, the validity of a
continuous approach becomes ever more tenuous.
As such, the adequacy of the law of mass action has been
questioned for describing intracellular reactions.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background: Law of Mass Action
The law of mass action considers chemical reactions to be
macroscopic under convective or diusive stirring, continuous
and deterministic.
These are evidently simplications, as it is well understood
that chemical reactions involve discrete, random collisions
between individual molecules.
As we consider smaller and smaller systems, the validity of a
continuous approach becomes ever more tenuous.
As such, the adequacy of the law of mass action has been
questioned for describing intracellular reactions.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background: Law of Mass Action
The law of mass action considers chemical reactions to be
macroscopic under convective or diusive stirring, continuous
and deterministic.
These are evidently simplications, as it is well understood
that chemical reactions involve discrete, random collisions
between individual molecules.
As we consider smaller and smaller systems, the validity of a
continuous approach becomes ever more tenuous.
As such, the adequacy of the law of mass action has been
questioned for describing intracellular reactions.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background: Law of Mass Action
The law of mass action considers chemical reactions to be
macroscopic under convective or diusive stirring, continuous
and deterministic.
These are evidently simplications, as it is well understood
that chemical reactions involve discrete, random collisions
between individual molecules.
As we consider smaller and smaller systems, the validity of a
continuous approach becomes ever more tenuous.
As such, the adequacy of the law of mass action has been
questioned for describing intracellular reactions.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background: Application of Stochastic Models
Arguments for the application of stochastic models for chemical
reactions come from at least three directions, since the models:
1
take into consideration the discrete character of the quantity
of components and the inherently random character of the
phenomena;
2
are in accordance with the theories of thermodynamics and
stochastic processes; and
3
are appropriate to describe small systems and instability
phenomena.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background: Application of Stochastic Models
Arguments for the application of stochastic models for chemical
reactions come from at least three directions, since the models:
1
take into consideration the discrete character of the quantity
of components and the inherently random character of the
phenomena;
2
are in accordance with the theories of thermodynamics and
stochastic processes; and
3
are appropriate to describe small systems and instability
phenomena.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background: Application of Stochastic Models
Arguments for the application of stochastic models for chemical
reactions come from at least three directions, since the models:
1
take into consideration the discrete character of the quantity
of components and the inherently random character of the
phenomena;
2
are in accordance with the theories of thermodynamics and
stochastic processes; and
3
are appropriate to describe small systems and instability
phenomena.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background: Simulation
Stochastic simulation methods
Nothing new?
Not just discrete-event simulation
Specialist method well-suited to large-scale systems
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background: Simulation
Stochastic simulation methods
Nothing new?
Not just discrete-event simulation
Specialist method well-suited to large-scale systems
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background: Simulation
Stochastic simulation methods
Nothing new?
Not just discrete-event simulation
Specialist method well-suited to large-scale systems
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Background: Simulation
Stochastic simulation methods
Nothing new?
Not just discrete-event simulation
Specialist method well-suited to large-scale systems
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Acknowledgements
H. Bolouri, J.T. Bradley, J. Bruck, K. Burrage, M. Calder, Y. Cao,
K.-H. Cho, A.J. Duguid, C. van Gend, M.A. Gibson, D.T. Gillespie,
J. Hillston, M. Khammash, W. Kolch, U. Kummer, D. Orrell,
L. Petzold, S. Ramsey, H.E. Samad, S. Schnell, N.T. Thomas,
T.E. Turner, M. Ullah, O. Wolkenhauer
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Outline
1
The deterministic and stochastic approaches
2
Stochastic simulation algorithms
3
Comparing stochastic simulation and ODEs
4
Modelling challenges
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Outline
1
The deterministic and stochastic approaches
2
Stochastic simulation algorithms
3
Comparing stochastic simulation and ODEs
4
Modelling challenges
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Deterministic: The law of mass action
The fundamental empirical law governing reaction rates in
biochemistry is the law of mass action.
This states that for a reaction in a homogeneous, free medium, the
reaction rate will be proportional to the concentrations of the
individual reactants involved.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Deterministic: The law of mass action
The fundamental empirical law governing reaction rates in
biochemistry is the law of mass action.
This states that for a reaction in a homogeneous, free medium, the
reaction rate will be proportional to the concentrations of the
individual reactants involved.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Deterministic: Michaelis-Menten kinetics
Consider the simple Michaelis-Menten reaction
S + E
k
1

k
1
C
k
2

E + P
For example, we have
dC
dt
= k
1
SE (k
1
+ k
2
)C
Hence, we can express any chemical system as a collection of
coupled non-linear rst order dierential equations.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Deterministic: Michaelis-Menten kinetics
Consider the simple Michaelis-Menten reaction
S + E
k
1

k
1
C
k
2

E + P
For example, we have
dC
dt
= k
1
SE (k
1
+ k
2
)C
Hence, we can express any chemical system as a collection of
coupled non-linear rst order dierential equations.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Deterministic: Michaelis-Menten kinetics
Consider the simple Michaelis-Menten reaction
S + E
k
1

k
1
C
k
2

E + P
For example, we have
dC
dt
= k
1
SE (k
1
+ k
2
)C
Hence, we can express any chemical system as a collection of
coupled non-linear rst order dierential equations.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Random processes
Whereas the deterministic approach outlined above is
essentially an empirical law, derived from in vitro experiments,
the stochastic approach is far more physically rigorous.
Fundamental to the principle of stochastic modelling is the
idea that molecular reactions are essentially random processes;
it is impossible to say with complete certainty the time at
which the next reaction within a volume will occur.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Random processes
Whereas the deterministic approach outlined above is
essentially an empirical law, derived from in vitro experiments,
the stochastic approach is far more physically rigorous.
Fundamental to the principle of stochastic modelling is the
idea that molecular reactions are essentially random processes;
it is impossible to say with complete certainty the time at
which the next reaction within a volume will occur.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Random processes
Whereas the deterministic approach outlined above is
essentially an empirical law, derived from in vitro experiments,
the stochastic approach is far more physically rigorous.
Fundamental to the principle of stochastic modelling is the
idea that molecular reactions are essentially random processes;
it is impossible to say with complete certainty the time at
which the next reaction within a volume will occur.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Predictability of macroscopic states
In macroscopic systems, with a large number of interacting
molecules, the randomness of this behaviour averages out so
that the overall macroscopic state of the system becomes
highly predictable.
It is this property of large scale random systems that enables a
deterministic approach to be adopted; however, the validity of
this assumption becomes strained in in vivo conditions as we
examine small-scale cellular reaction environments with
limited reactant populations.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Predictability of macroscopic states
In macroscopic systems, with a large number of interacting
molecules, the randomness of this behaviour averages out so
that the overall macroscopic state of the system becomes
highly predictable.
It is this property of large scale random systems that enables a
deterministic approach to be adopted; however, the validity of
this assumption becomes strained in in vivo conditions as we
examine small-scale cellular reaction environments with
limited reactant populations.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Predictability of macroscopic states
In macroscopic systems, with a large number of interacting
molecules, the randomness of this behaviour averages out so
that the overall macroscopic state of the system becomes
highly predictable.
It is this property of large scale random systems that enables a
deterministic approach to be adopted; however, the validity of
this assumption becomes strained in in vivo conditions as we
examine small-scale cellular reaction environments with
limited reactant populations.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Propensity function
As explicitly derived by Gillespie, the stochastic model uses basic
Newtonian physics and thermodynamics to arrive at a form often
termed the propensity function that gives the probability a

of
reaction occurring in time interval (t, t + dt).
a

dt = h

dt
where the M reaction mechanisms are given an arbitrary index
(1 M), h

denotes the number of possible combinations of

reactant molecules involved in reaction , and c

is a stochastic
rate constant.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Fundamental hypothesis
The rate constant c

is dependent on the radii of the molecules

involved in the reaction, and their average relative velocities a
property that is itself a direct function of the temperature of the
system and the individual molecular masses.
These quantities are basic chemical properties which for most
systems are either well known or easily measurable. Thus, for a
given chemical system, the propensity functions, a

can be easily
determined.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Fundamental hypothesis
The rate constant c

is dependent on the radii of the molecules

involved in the reaction, and their average relative velocities a
property that is itself a direct function of the temperature of the
system and the individual molecular masses.
These quantities are basic chemical properties which for most
systems are either well known or easily measurable. Thus, for a
given chemical system, the propensity functions, a

can be easily
determined.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Grand probability function
The stochastic formulation proceeds by considering the grand
probability function Pr(X; t) probability that there will be
present in the volume V at time t, X
i
of species S
i
, where
X (X
1
, X
2
, . . . X
N
) is a vector of molecular species populations.
Evidently, knowledge of this function provides a complete
understanding of the probability distribution of all possible states
at all times.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Grand probability function
The stochastic formulation proceeds by considering the grand
probability function Pr(X; t) probability that there will be
present in the volume V at time t, X
i
of species S
i
, where
X (X
1
, X
2
, . . . X
N
) is a vector of molecular species populations.
Evidently, knowledge of this function provides a complete
understanding of the probability distribution of all possible states
at all times.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Innitesimal time interval
By considering a discrete innitesimal time interval (t, t + dt) in
which either 0 or 1 reactions occur we see that there exist only
M + 1 distinct congurations at time t that can lead to the state
X at time t + dt.
Pr(X; t + dt)
= Pr(X; t) Pr(no state change over dt)
+

M
=1
Pr(X v

; t) Pr(state change to X over dt)

where v

is a stoichiometric vector dening the result of reaction

on state vector X, i.e. X X + v

after an occurrence of
reaction .
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Innitesimal time interval
By considering a discrete innitesimal time interval (t, t + dt) in
which either 0 or 1 reactions occur we see that there exist only
M + 1 distinct congurations at time t that can lead to the state
X at time t + dt.
Pr(X; t + dt)
= Pr(X; t) Pr(no state change over dt)
+

M
=1
Pr(X v

; t) Pr(state change to X over dt)

where v

is a stoichiometric vector dening the result of reaction

on state vector X, i.e. X X + v

after an occurrence of
reaction .
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Innitesimal time interval
By considering a discrete innitesimal time interval (t, t + dt) in
which either 0 or 1 reactions occur we see that there exist only
M + 1 distinct congurations at time t that can lead to the state
X at time t + dt.
Pr(X; t + dt)
= Pr(X; t) Pr(no state change over dt)
+

M
=1
Pr(X v

; t) Pr(state change to X over dt)

where v

is a stoichiometric vector dening the result of reaction

on state vector X, i.e. X X + v

after an occurrence of
reaction .
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Innitesimal time interval
By considering a discrete innitesimal time interval (t, t + dt) in
which either 0 or 1 reactions occur we see that there exist only
M + 1 distinct congurations at time t that can lead to the state
X at time t + dt.
Pr(X; t + dt)
= Pr(X; t) Pr(no state change over dt)
+

M
=1
Pr(X v

; t) Pr(state change to X over dt)

where v

is a stoichiometric vector dening the result of reaction

on state vector X, i.e. X X + v

after an occurrence of
reaction .
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: State change probabilities
Pr(no state change over dt)
1
M

=1
a

(X)dt
Pr(state change to X over dt)
M

=1
Pr(X v

; t)a

(X v

)dt
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: State change probabilities
Pr(no state change over dt)
1
M

=1
a

(X)dt
Pr(state change to X over dt)
M

=1
Pr(X v

; t)a

(X v

)dt
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Partial derivatives
Pr(X; t)
t
= lim
dt0
Pr(X; t + dt) Pr(X; t)
dt
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Chemical Master Equation
Applying this, and re-arranging the former, leads us to an
important partial dierential equation (PDE) known as the
Chemical Master Equation (CME).
Pr(X; t)
t
=
M

=1
a

(X v

) Pr(X v

; t) a

(X) Pr(X; t)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Chemical Master Equation
Applying this, and re-arranging the former, leads us to an
important partial dierential equation (PDE) known as the
Chemical Master Equation (CME).
Pr(X; t)
t
=
M

=1
a

(X v

) Pr(X v

; t) a

(X) Pr(X; t)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Chemical Master Equation
Applying this, and re-arranging the former, leads us to an
important partial dierential equation (PDE) known as the
Chemical Master Equation (CME).
Pr(X; t)
t
=
M

=1
a

(X v

) Pr(X v

; t) a

(X) Pr(X; t)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic: Chemical Master Equation
Applying this, and re-arranging the former, leads us to an
important partial dierential equation (PDE) known as the
Chemical Master Equation (CME).
Pr(X; t)
t
=
M

=1
a

(X v

) Pr(X v

; t) a

(X) Pr(X; t)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
The problem with the Chemical Master Equation
The CME is really a set of nearly as many coupled ordinary
dierential equations as there are combinations of molecules
that can exist in the system!
The CME can be solved analytically for only a very few very
simple systems, and numerical solutions are usually
prohibitively dicult.
D. Gillespie and L. Petzold.
chapter Numerical Simulation for Biochemical Kinetics, in System
Modelling in Cellular Biology, editors Z. Szallasi, J. Stelling and V.
Periwal.
MIT Press, 2006.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Outline
1
The deterministic and stochastic approaches
2
Stochastic simulation algorithms
3
Comparing stochastic simulation and ODEs
4
Modelling challenges
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Breakthrough: Gillespies Stochastic simulation algorithms
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Biography: Daniel T. Gillespie
1960 BA from Rice University
1968 PhD from Johns Hopkins University
19681971 Postdoc at the University of Marylands Institute for
Molecular Physics.
19712001 Research Physicist in the Earth & Planetary Sciences
Division of the Naval Air Warfare Center in China
Lake, California.
2001 Retirement from Civil Service. Begins consultancy for
California Institute of Technology and the Molecular
Sciences Institute, working mostly with Linda Petzold
and her group at the University of California at Santa
Barbara.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Books by Daniel T. Gillespie
A Quantum Mechanics Primer (1970)
Markov Processes: An Introduction for Physical Scientists
(1992)
Biography of radio comedy writer Tom Koch (2004)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic simulation algorithms
Gillespies Stochastic Simulation Algorithm (SSA) is essentially an
exact procedure for numerically simulating the time evolution of a
well-stirred chemically reacting system by taking proper account of
the randomness inherent in such a system.
It is rigorously based on the same microphysical premise that
underlies the chemical master equation and gives a more realistic
representation of a systems evolution than the deterministic
reaction rate equation (RRE) represented mathematically by ODEs.
As with the chemical master equation, the SSA converges, in the
limit of large numbers of reactants, to the same solution as the law
of mass action.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic simulation algorithms
Gillespies Stochastic Simulation Algorithm (SSA) is essentially an
exact procedure for numerically simulating the time evolution of a
well-stirred chemically reacting system by taking proper account of
the randomness inherent in such a system.
It is rigorously based on the same microphysical premise that
underlies the chemical master equation and gives a more realistic
representation of a systems evolution than the deterministic
reaction rate equation (RRE) represented mathematically by ODEs.
As with the chemical master equation, the SSA converges, in the
limit of large numbers of reactants, to the same solution as the law
of mass action.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic simulation algorithms
Gillespies Stochastic Simulation Algorithm (SSA) is essentially an
exact procedure for numerically simulating the time evolution of a
well-stirred chemically reacting system by taking proper account of
the randomness inherent in such a system.
It is rigorously based on the same microphysical premise that
underlies the chemical master equation and gives a more realistic
representation of a systems evolution than the deterministic
reaction rate equation (RRE) represented mathematically by ODEs.
As with the chemical master equation, the SSA converges, in the
limit of large numbers of reactants, to the same solution as the law
of mass action.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies exact SSA (1977)
The algorithm takes time steps of variable length, based on
the rate constants and population size of each chemical
species.
The probability of one reaction occurring relative to another is
dictated by their relative propensity functions.
According to the correct probability distribution derived from
the statistical thermodynamics theory, a random variable is
then used to choose which reaction will occur, and another
random variable determines how long the step will last.
The chemical populations are altered according to the
stoichiometry of the reaction and the process is repeated.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies exact SSA (1977)
The algorithm takes time steps of variable length, based on
the rate constants and population size of each chemical
species.
The probability of one reaction occurring relative to another is
dictated by their relative propensity functions.
According to the correct probability distribution derived from
the statistical thermodynamics theory, a random variable is
then used to choose which reaction will occur, and another
random variable determines how long the step will last.
The chemical populations are altered according to the
stoichiometry of the reaction and the process is repeated.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies exact SSA (1977)
The algorithm takes time steps of variable length, based on
the rate constants and population size of each chemical
species.
The probability of one reaction occurring relative to another is
dictated by their relative propensity functions.
According to the correct probability distribution derived from
the statistical thermodynamics theory, a random variable is
then used to choose which reaction will occur, and another
random variable determines how long the step will last.
The chemical populations are altered according to the
stoichiometry of the reaction and the process is repeated.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies exact SSA (1977)
The algorithm takes time steps of variable length, based on
the rate constants and population size of each chemical
species.
The probability of one reaction occurring relative to another is
dictated by their relative propensity functions.
According to the correct probability distribution derived from
the statistical thermodynamics theory, a random variable is
then used to choose which reaction will occur, and another
random variable determines how long the step will last.
The chemical populations are altered according to the
stoichiometry of the reaction and the process is repeated.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic simulation: Job done!
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic simulation: realisations and ensembles
The SSA computes one realisation of a dynamic trajectory of a
chemically reacting system. Often an ensemble of trajectories is
computed, to obtain an estimate of the probability density function
of the system.
The dynamic evolution of the probability density function is given
by the Chemical Master Equation.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies SSA is a Monte Carlo Markov Chain simulation
The SSA is a Monte Carlo type method. With the SSA one may
approximate any variable of interest by generating many
trajectories and observing the statistics of the values of the
variable. Since many trajectories are needed to obtain a reasonable
approximation, the eciency of the SSA is of critical importance.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Computational cost of Gillespies exact algorithm
The cost of this detailed stochastic simulation algorithm is the
likely large amounts of computing time.
The key issue is that the time step for the next reaction can be
very small indeed if we are to guarantee that only one reaction can
take place in a given time interval.
Increasing the molecular population or number of reaction
mechanisms necessarily requires a corresponding decrease in the
time interval. The SSA can be very computationally inecient
especially when there are large numbers of molecules or the
propensity functions are large.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Computational cost of Gillespies exact algorithm
The cost of this detailed stochastic simulation algorithm is the
likely large amounts of computing time.
The key issue is that the time step for the next reaction can be
very small indeed if we are to guarantee that only one reaction can
take place in a given time interval.
Increasing the molecular population or number of reaction
mechanisms necessarily requires a corresponding decrease in the
time interval. The SSA can be very computationally inecient
especially when there are large numbers of molecules or the
propensity functions are large.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gibson and Bruck (2000)
Gibson and Bruck rened the rst reaction SSA of Gillespie by
reducing the number of random variables that need to be
simulated.
This can be eective for systems in which some reactions occur
much more frequently than others.
M.A. Gibson and J. Bruck.
Ecient exact stochastic simulation of chemical systems with many
species and many channels.
J. Comp. Phys., 104:18761889, 2000.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Variants of SSA
Gillespie developed two dierent but equivalent formulations of the
SSA: the Direct Method (DM) and the First Reaction Method
(FRM). A third formulation of the SSA is the Next Reaction
Method (NRM) of Gibson and Bruck. The NRM can be viewed as
an extension of the FRM, but it is much more ecient than the
latter.
It was widely believed that Gibson and Brucks method (the Next
Reaction Method) was more ecient than Gillespies Direct
Method (DM). This conclusion is based on a count of arithmetic
operations.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Variants of SSA
Gillespie developed two dierent but equivalent formulations of the
SSA: the Direct Method (DM) and the First Reaction Method
(FRM). A third formulation of the SSA is the Next Reaction
Method (NRM) of Gibson and Bruck. The NRM can be viewed as
an extension of the FRM, but it is much more ecient than the
latter.
It was widely believed that Gibson and Brucks method (the Next
Reaction Method) was more ecient than Gillespies Direct
Method (DM). This conclusion is based on a count of arithmetic
operations.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gibson and Bruck challenged (2004)
It was established by Cao, Li and Petzold (2004) that Gibson and
Brucks analysis misses the dominant cost of the NRM, which is
maintaining the priority queue data structure of the tentative
reaction times and that good implementations of DM such as the
Optimised Direct Method (ODM) have lower asymptotic
complexity than Gibson and Brucks method.
Y. Cao, H. Li, and L. Petzold.
Ecient formulation of the stochastic simulation algorithm for
chemically reacting systems.
J. Chem. Phys, 121(9):40594067, 2004.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Enhanced stochastic simulation techniques
If the system under study possesses a macroscopically innitesimal
timescale so that during any dt all of the reaction channels can re
many times, yet none of the propensity functions change
appreciably, then the discrete Markov process as described by the
SSA can be approximated by a continuous Markov process.
This Markov process is described by the Chemical Langevin
Equation (CLE), which is a stochastic ordinary dierential equation
(SDE).
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Enhanced stochastic simulation techniques
If the system under study possesses a macroscopically innitesimal
timescale so that during any dt all of the reaction channels can re
many times, yet none of the propensity functions change
appreciably, then the discrete Markov process as described by the
SSA can be approximated by a continuous Markov process.
This Markov process is described by the Chemical Langevin
Equation (CLE), which is a stochastic ordinary dierential equation
(SDE).
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Enhanced stochastic simulation techniques
If the system under study possesses a macroscopically innitesimal
timescale so that during any dt all of the reaction channels can re
many times, yet none of the propensity functions change
appreciably, then the discrete Markov process as described by the
SSA can be approximated by a continuous Markov process.
This Markov process is described by the Chemical Langevin
Equation (CLE), which is a stochastic ordinary dierential equation
(SDE).
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Enhanced stochastic simulation techniques
If the system under study possesses a macroscopically innitesimal
timescale so that during any dt all of the reaction channels can re
many times, yet none of the propensity functions change
appreciably, then the discrete Markov process as described by the
SSA can be approximated by a continuous Markov process.
This Markov process is described by the Chemical Langevin
Equation (CLE), which is a stochastic ordinary dierential equation
(SDE).
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Enhanced stochastic simulation techniques
If the system under study possesses a macroscopically innitesimal
timescale so that during any dt all of the reaction channels can re
many times, yet none of the propensity functions change
appreciably, then the discrete Markov process as described by the
SSA can be approximated by a continuous Markov process.
This Markov process is described by the Chemical Langevin
Equation (CLE), which is a stochastic ordinary dierential equation
(SDE).
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic Dierential Equations
A stochastic dierential equation (SDE)
dX
t
= a(t, X
t
)dt + b(t, X
t
)dW
t
is interpreted as a stochastic integral equation
X
t
= X
t
0
+

t
t
0
a(s, X
s
)ds +

t
t
0
b(s, X
s
)dW
s
where the rst integral is a Lebesgue (or Riemann) integral for
each sample path and the second integral is usually an Ito integral.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Chemical Langevin Equation
The Langevin equation
dX
t
= aX
t
dt + dW
t
is a linear SDE with additive noise. The solution for t
0
= 0 is
X
t
= X
0
e
at
+ e
at

t
0
e
as
dW
s
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies tau-leap method (2001)
Gillespie proposed two new methods, namely the -leap method
and the midpoint -leap method in order to improve the eciency
of the SSA while maintaining acceptable losses in accuracy.
Daniel T. Gillespie.
Approximate accelerated stochastic simulation of chemically reacting
systems.
J. Comp. Phys., 115(4):17161733, 2001.
The key idea here is to take a larger time step and allow for more
reactions to take place in that step, but under the proviso that the
propensity functions do not change too much in that interval. By
means of a Poisson approximation, the tau-leaping method can
leap over many reactions.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies tau-leap method (2001)
Gillespie proposed two new methods, namely the -leap method
and the midpoint -leap method in order to improve the eciency
of the SSA while maintaining acceptable losses in accuracy.
Daniel T. Gillespie.
Approximate accelerated stochastic simulation of chemically reacting
systems.
J. Comp. Phys., 115(4):17161733, 2001.
The key idea here is to take a larger time step and allow for more
reactions to take place in that step, but under the proviso that the
propensity functions do not change too much in that interval. By
means of a Poisson approximation, the tau-leaping method can
leap over many reactions.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies tau-leap method (signicance)
For many problems, the tau-leaping method can approximate the
stochastic behaviour of the system very well.
The tau-leaping method connects the SSA in the discrete
stochastic regime to the explicit Euler method for the chemical
Langevin equation in the continuous stochastic regime and the
RRE in the continuous deterministic regime.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies tau-leap method (signicance)
For many problems, the tau-leaping method can approximate the
stochastic behaviour of the system very well.
The tau-leaping method connects the SSA in the discrete
stochastic regime to the explicit Euler method for the chemical
Langevin equation in the continuous stochastic regime and the
RRE in the continuous deterministic regime.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies tau-leap method (signicance)
For many problems, the tau-leaping method can approximate the
stochastic behaviour of the system very well.
The tau-leaping method connects the SSA in the discrete
stochastic regime to the explicit Euler method for the chemical
Langevin equation in the continuous stochastic regime and the
RRE in the continuous deterministic regime.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies tau-leap method (signicance)
For many problems, the tau-leaping method can approximate the
stochastic behaviour of the system very well.
The tau-leaping method connects the SSA in the discrete
stochastic regime to the explicit Euler method for the chemical
Langevin equation in the continuous stochastic regime and the
RRE in the continuous deterministic regime.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies tau-leap method (signicance)
For many problems, the tau-leaping method can approximate the
stochastic behaviour of the system very well.
The tau-leaping method connects the SSA in the discrete
stochastic regime to the explicit Euler method for the chemical
Langevin equation in the continuous stochastic regime and the
RRE in the continuous deterministic regime.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies tau-leap method (drawback)
The use of approximation in Poisson methods leads to the
possibility of negative molecular numbers being predicted
something with no physical explanation.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies Modied Poisson tau-leap methods (2005)
Gillespies modied Poisson tau-leaping method introduces a
second control parameter whose value dials the procedure from the
original Poisson tau-leaping method at one extreme to the exact
SSA at the other.
Any reaction channel with a positive propensity function which is
within n
c
rings of exhausting its reactants is termed a critical
reaction.
Y. Cao, D. Gillespie, and L. Petzold.
Avoiding negative populations in explicit tau leaping.
J. Chem. Phys, 123(054104), 2005.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies Modied Poisson tau-leap methods (2005)
Gillespies modied Poisson tau-leaping method introduces a
second control parameter whose value dials the procedure from the
original Poisson tau-leaping method at one extreme to the exact
SSA at the other.
Any reaction channel with a positive propensity function which is
within n
c
rings of exhausting its reactants is termed a critical
reaction.
Y. Cao, D. Gillespie, and L. Petzold.
Avoiding negative populations in explicit tau leaping.
J. Chem. Phys, 123(054104), 2005.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies Modied Poisson tau-leap methods (2006)
The modied algorithm chooses in such a way that no more than
one ring of all the critical reactions can occur during the leap.
The probability of producing a negative population is reduced to
nearly zero.
If a negative population does occur the leap can simply be rejected
and repeated with reduced by half, or the entire simulation can
be abandoned and repeated for larger n
c
.
Y. Cao, D. Gillespie, and L. Petzold.
Ecient stepsize selection for the tau-leaping method.
J. Chem. Phys, 2006.
To appear.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies Modied Poisson tau-leap methods (2006)
The modied algorithm chooses in such a way that no more than
one ring of all the critical reactions can occur during the leap.
The probability of producing a negative population is reduced to
nearly zero.
If a negative population does occur the leap can simply be rejected
and repeated with reduced by half, or the entire simulation can
be abandoned and repeated for larger n
c
.
Y. Cao, D. Gillespie, and L. Petzold.
Ecient stepsize selection for the tau-leaping method.
J. Chem. Phys, 2006.
To appear.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Family of stochastic simulation algorithms
FASTEST, BEST
Discrete, exact Continuous, approximate
Modied Poisson leap (2005)
leap (2001)
Logarithmic Direct Method (2006)
Sorting Direct Method (2005)
Optimised Direct Method (2004)
Next Reaction Method (2000)
Direct Method (1977)
First Reaction Method (1977)
SLOWEST, WORST
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Outline
1
The deterministic and stochastic approaches
2
Stochastic simulation algorithms
3
Comparing stochastic simulation and ODEs
4
Modelling challenges
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Comparing stochastic simulation and ODEs
We know that stochastic simulation can allow us to observe
phenomena which ODEs cannot.
Are there places where they agree?
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Comparing stochastic simulation and ODEs
We know that stochastic simulation can allow us to observe
phenomena which ODEs cannot.
Are there places where they agree?
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
A simple example: processors and resources
Proc
0
def
= (task1, ).Proc
1
Proc
1
def
= (task2, r
2
).Proc
0
Res
0
def
= (task1, r
1
).Res
1
Res
1
def
= (reset, s).Res
0
Proc
0
[P]

{task1}
Res
0
[R]
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
A simple example: processors and resources
Proc
0
def
= (task1, ).Proc
1
Proc
1
def
= (task2, r
2
).Proc
0
Res
0
def
= (task1, r
1
).Res
1
Res
1
def
= (reset, s).Res
0
Proc
0
[P]

{task1}
Res
0
[R]
CTMC interpretation
Processors (P) Resources (R) States (2
P+R
)
1 1 4
2 1 8
2 2 16
3 2 32
3 3 64
4 3 128
4 4 256
5 4 512
5 5 1024
6 5 2048
6 6 4096
7 6 8192
7 7 16384
8 7 32768
8 8 65536
9 8 131072
9 9 262144
10 9 524288
10 10 1048576
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
A simple example: processors and resources
Proc
0
def
= (task1, ).Proc
1
Proc
1
def
= (task2, r
2
).Proc
0
Res
0
def
= (task1, r
1
).Res
1
Res
1
def
= (reset, s).Res
0
Proc
0
[P]

{task1}
Res
0
[R]
ODE interpretation
dProc
0
dt
= r
1
min(Proc
0
, Res
0
)
+r
2
Proc
1
dProc
1
dt
= r
1
min(Proc
0
, Res
0
)
r
2
Proc
1
dRes
0
dt
= r
1
min(Proc
0
, Res
0
)
+s Res
1
dRes
1
dt
= r
1
min(Proc
0
, Res
0
)
s Res
1
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources (SSA run A)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources (SSA run B)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources (SSA run C)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources (SSA run D)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources (average of 10 runs)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources (average of 100 runs)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources (average of 1000 runs)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources (average of 10000 runs)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources (ODE solution)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
From realisations to ensembles
As we repeatedly sample from the underlying random number
distributions the average of the samples tends to the mean.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources: scaling out
What eect does increasing the number of copies have?
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources (single SSA run, 100/80)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources (single SSA run, 1,000/800)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources (single SSA run, 10,000/8,000)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Processors and resources (single SSA run, 100,000/80,000)
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
From the microscopic to the macroscopic domain
Each specic run is individually in closer and closer agreement with
the deterministic approach as the number of molecules in the
system increases.
This is a direct eect of the inherent averaging of macroscopic
properties of a system of many particles.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Conclusions from the comparison
1
These results provide clear verication of the compatibility of
the deterministic and stochastic approaches.
2
They also illustrate the validity of the deterministic approach
in systems containing as few as 100 copies of components.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Outline
1
The deterministic and stochastic approaches
2
Stochastic simulation algorithms
3
Comparing stochastic simulation and ODEs
4
Modelling challenges
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Modelling challenges: stiness
A problem for modelling temporal evolution is stiness. Some
reactions are much faster than others and quickly reach a stable
state. The dynamics of the system is driven by the slow reactions.
Most chemical systems, whether considered at a scale appropriate
to stochastic or to deterministic simulation, involve several widely
varying time scales, so such systems are nearly always sti.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Modelling challenges: stiness
A problem for modelling temporal evolution is stiness. Some
reactions are much faster than others and quickly reach a stable
state. The dynamics of the system is driven by the slow reactions.
Most chemical systems, whether considered at a scale appropriate
to stochastic or to deterministic simulation, involve several widely
varying time scales, so such systems are nearly always sti.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Modelling challenges: stiness
A problem for modelling temporal evolution is stiness. Some
reactions are much faster than others and quickly reach a stable
state. The dynamics of the system is driven by the slow reactions.
Most chemical systems, whether considered at a scale appropriate
to stochastic or to deterministic simulation, involve several widely
varying time scales, so such systems are nearly always sti.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Modelling challenges: multiscale populations
The multiscale population problem arises when some species are
present in relatively small quantities and should be modelled by a
discrete stochastic process, whereas other species are present in
larger quantities and are more eciently modelled by a
deterministic ordinary dierential equation (or at some scale in
between). SSA treats all of the species as discrete stochastic
processes.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies multiscale SSA methods (2005)
SSA is used for slow reactions or species with small populations.
The multiscale SSA method generalizes this idea to the case in
which species with small population are involved in fast reactions.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies slow-scale SSA methods (2005)
The setting for Gillespies slow-scale SSA method is
S + E
k
1

k
1
C
k
2

E + P
where
k
1
k
2
Slow-scale SSA explicitly simulates only the relatively rare
conversion reactions, skipping over occurrences of the other two
less interesting but much more frequent reactions.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Gillespies slow-scale SSA methods (2005)
The setting for Gillespies slow-scale SSA method is
S + E
k
1

k
1
C
k
2

E + P
where
k
1
k
2
Slow-scale SSA explicitly simulates only the relatively rare
conversion reactions, skipping over occurrences of the other two
less interesting but much more frequent reactions.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Comparing SSA and Slow-Scale SSA results
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Comparing SSA and Slow-Scale SSA results
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Conclusions
Stochastic simulation is a well-founded method for simulating
in vivo reactions.
Gillespies SSA can be more accurate than ODEs at low
molecular numbers; compatible with them at large molecular
numbers.
Recent explosion of interest in the subject with many new
variants of the SSA algorithm.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Excellent introductory papers
T.E. Turner, S. Schnell, and K. Burrage.
Stochastic approaches for modelling in vivo reactions.
Computational Biology and Chemistry, 28:165178, 2004.
D. Gillespie and L. Petzold.
System Modelling in Cellular Biology, chapter Numerical Simulation
for Biochemical Kinetics,.
MIT Press, 2006.
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation
The deterministic and stochastic approaches
Stochastic simulation algorithms
Comparing stochastic simulation and ODEs
Modelling challenges
Stochastic simulation software
S. Ramsey, D. Orrell, and H. Bolouri.
Dizzy: stochastic simulation of large-scale genetic regulatory
networks.
J. Bioinf. Comp. Biol., 3(2):415436, 2005.
http://magnet.systemsbiology.net/software/Dizzy
Stephen Gilmore. Informatics, University of Edinburgh. An Introduction to Stochastic Simulation