In this paragraph, we will discuss the implementation of
+
the agent-based module. First, the problem of computing
+
pairwise interactions is discussed and a nearest-neighbor
+
algorithm is introduced to reduce the computation time.
+
Then the choice of a timestep is considered especially
+
with regards to the cell-cell interaction scheme. Finally,
+
the boundary conditions for agents are defined and explained.
</p>
</p>
</div><!-- part -->
</div><!-- part -->
Line 1,706:
Line 1,712:
<p>
<p>
The cell-cell interactions have already been fully described in the paragraphs above. However, solving the equation of motion of an agent in its current form requires the computation of the force due to every other agent. If we take N to be the number of agents, that means $N*(N-1)$ amount of force computations are needed in total and therefore the computation time grows as $O(N^2)$. To put this in perspective, if we simulate a thousand agents, the amount of interactions adds up to around one million. This puts a heavy computational load on the model which can be reduced greatly by using a smarter algorithm. The crucial observation here is that the force goes to zero when the distance between two cells is larger than $2\cdot r_{cutoff}$. An agent therefore only interacts with its nearest neighbors, meaning that the naive implementation wastes a lot of time computing interactions which have no effect anyways. Moreover, as the simulation progresses, an agent’s neighbors do not vary greatly from one timestep to another. A more efficient approach then consists of periodically searching for the nearest neighbors within a fixed radius, storing them in a list for every agent and updating their coordinates for all intermediate timesteps.
The cell-cell interactions have already been fully described in the paragraphs above. However, solving the equation of motion of an agent in its current form requires the computation of the force due to every other agent. If we take N to be the number of agents, that means $N*(N-1)$ amount of force computations are needed in total and therefore the computation time grows as $O(N^2)$. To put this in perspective, if we simulate a thousand agents, the amount of interactions adds up to around one million. This puts a heavy computational load on the model which can be reduced greatly by using a smarter algorithm. The crucial observation here is that the force goes to zero when the distance between two cells is larger than $2\cdot r_{cutoff}$. An agent therefore only interacts with its nearest neighbors, meaning that the naive implementation wastes a lot of time computing interactions which have no effect anyways. Moreover, as the simulation progresses, an agent’s neighbors do not vary greatly from one timestep to another. A more efficient approach then consists of periodically searching for the nearest neighbors within a fixed radius, storing them in a list for every agent and updating their coordinates for all intermediate timesteps.
−
+
<br/>
+
<br/>
Since the agents are programmed to repel each other when they approach each other too closely, they will evolve to a rather uniform distribution. The most appropriate fixed-radius nearest neighbor search algorithm in that case is the cell technique. In this algorithm, the agents are structured by placing a rectangular grid of cells over the domain and assigning every agent to a cell (fig9). Determining which cell an agent belongs to is easily done by rounding down the x and y-coordinates to the nearest multiple of the grid spacing. If the grid spacing is chosen so that interactions do not reach further than adjacent cells, every agent only needs to look for neighbors in 9 cells (its own cells plus 8 adjacent cells) instead of the entire domain.
Since the agents are programmed to repel each other when they approach each other too closely, they will evolve to a rather uniform distribution. The most appropriate fixed-radius nearest neighbor search algorithm in that case is the cell technique. In this algorithm, the agents are structured by placing a rectangular grid of cells over the domain and assigning every agent to a cell (fig9). Determining which cell an agent belongs to is easily done by rounding down the x and y-coordinates to the nearest multiple of the grid spacing. If the grid spacing is chosen so that interactions do not reach further than adjacent cells, every agent only needs to look for neighbors in 9 cells (its own cells plus 8 adjacent cells) instead of the entire domain.
Line 1,960:
Line 1,967:
<div id="toggleten">
<div id="toggleten">
<p>
<p>
−
As mentioned earlier the concentrations of AHL and Leucine are modeled using partial differential equations.
+
As mentioned earlier the concentrations of AHL and Leucine
−
In the colony level model these equations are solved explicitly. Explicit schemes do not require a lot of
+
are modeled using partial differential equations.
−
work per time step, but unfortunately are not unconditionally stable. In two dimensions the grid ratios
+
In the colony level model these equations are solved
−
$dt/dx^2$ and $dt/dy^2$ can not exceed $dt/dx^2 + dt/dy^2 \leq \frac{1}{2}$ for the solver to be stable.
+
explicitly. Explicit schemes do not require a lot of
−
When computing the solution of the hybrid model this requrement forces us to spend a lot of CPU time solving
+
work per time step, but unfortunately are not
−
partial differential equations that could be better spent simulation the agents. Therefore an implicit ADI
+
unconditionally stable. In two dimensions the grid ratios
−
Alternating direction implicit scheme has been implemented. ADI-schemes are unconditionally stable, which
+
$dt/dx^2$ and $dt/dy^2$ can not exceed $dt/dx^2 + dt/dy^2
−
allows it to take large time steps with the PDE solver. We used the following scheme:
$C$: concentration [nmol/cl]
$\vec{r}$: position vector [µm]
$t$: time [h]
$D$: diffusion constant [µm^2/h]
$C$: concentration [nmol/cl]
$\vec{r}$: position vector [µm]
$t$: time [h]
$\alpha$: production constant [nmol/h]
$\rho_A$: density of type A bacteria [#/cl]
$C$: concentration [nmol/cl]
$\vec{r}$: position vector [µm]
$t$: time [h]
$k$: degradation constant [1/h]
$C$: concentration [nmol/cl]
$\vec{r}$: position vector [µm]
$t$: time [h]
$D$: diffusion constant [µm^2/h]
$\alpha$: production constant [nmol/h]
$\rho_A$: density of type A bacteria [#/cl]
$k$: degradation constant [1/h]
$\vec{r}$: position vector [µm]
$t$: time [h]
$\vec{F}_{applied}$: applied force [mN]
$\gamma$: frictional coefficient [mN*h/µm]
$\vec{r}$: position vector [µm]
$t$: time [h]
$\chi$: chemotactic sensitivity [µm^2*cl/(h*nmol)]
$S$: chemoattractant concentration [nmol/cl]
$\mu$: bacterial diffusion constant [µm^2/h]
$\vec{W}$: Wiener process [h^(1/2)]
$\kappa$: chemotactic sensitivity constant [-]
$n$: bacteria density [#/cl]
$\vec{r}$: position vector [µm]
$t$: time [h]
$\mu$: bacterial diffusion constant [µm^2/h]
$\chi$: chemotactic sensitivity [µm^2*cl/(h*nmol)]
$S$: chemoattractant concentration [nmol/cl]
$\kappa$: chemotactic sensitivity coefficient [-]
$\vec{F}_i$: cell-cell force acting on cell i [mN]
$E_p$: interaction potential [nJ]
$r_{ij}$: intercellular distance [µm]
$\vec{e}_{ij}$: unit vector from cell i to cell j [-]
$k$: “spring constant” [mN/µm]
$R_{cutoff}$: cutoff intercellular distance [µm]
$R_0$: equilibrium intercellular distance [µm]
$\vec{r}$: position vector [µm]
$t$: time [h]
$\mu$: bacterial diffusion constant [µm^2/h]
$H$: concentration of AHL [nmol/cl]
$\vec{W}$: Wiener process [h^(1/2)]
$\gamma$: frictional coefficient [mN*h/µm]
$E_p$: interaction potential [nJ]
$r_{ij}$: intercellular distance [µm]
$\vec{e}_{ij}$: unit vector from cell i to cell j
$\chi$: chemotactic sensitivity [µm^2*cl/(h*nmol)]
$L$: concentration of leucine [nmol/cl]
$\chi$: chemotactic sensitivity [µm^2*cl/(h*nmol)]
$L$: concentration of leucine [nmol/cl]
$H$: concentration of AHL [nmol/cl]
$\mu$: bacterial diffusion constant [µm^2/h]
$\kappa$: chemotactic sensitivity constant [-]
$\vec{r}$: position vector [µm]
$t$: time [h]
$\mu$: bacterial diffusion constant [µm^2/h]
$H$: concentration of AHL [nmol/cl]
$\vec{r}$: position vector [µm]
$t$: time [h]
$K$: probability density [-]
$x$: stochastic variable [-]
$K_h$: kernel density [#/µm]
$x$: position [µm]
$h$: bandwidth [µm]
$\rho$: agent density [#/cl]
$\beta$: conversion factor [µm/cl]
$N$: number of agents [#]
$K_h$: kernel density [#/µm]
$x$: position [µm]
$x_i$: position of ith agent [µm]
Conversion factor $\beta$ is needed to express the 1-D density [#/µm]
in units used in the PDE model [#/cl].
For the 2-D KDE $\beta$ has units [µm^2/cl]. The value of $\beta$
depends on assumptions made about the dimensions of the physical domain
which are not modeled. For example, when modeling a 2-D virtual domain
this constant will be high if bacteria are densely distributed
along the height of the experimental petri dish.
$f$: function, e.g. concentration [nmol/cl]
$\hat{f}$: interpolated function, e.g. [nmol/cl]
$x$: independent variable, e.g. position [µm]
$x_i$: independent variable for ith measurement, e.g. [µm]
$f$: function, e.g. concentration [nmol/cl]
$\hat{f}$: interpolated function, e.g. [nmol/cl]
$x$: independent variable, e.g. position [µm]
$x_i$: independent variable for measurement (i,j), e.g. [µm]
$y$: independent variable, e.g. position [µm]
$y_j$: independent variable for measurement (i,j), e.g. [µm]
$Q_{i,j}$: symbol for grid point corresponding to measurement (i,j)
$f$: function, e.g. concentration [nmol/cl]
$\hat{f}$: interpolated function, e.g. [nmol/cl]
$x$: independent variable, e.g. position [µm]
$x_i$: independent variable for measurement (i,j), e.g. [µm]
$y$: independent variable, e.g. position [µm]
$y_j$: independent variable for measurement (i,j), e.g. [µm]
$Q_{i,j}$: grid point corresponding to measurement (i,j) [-]
$Re$: Reynolds number [-]
$\rho$: mass density [kg/m^3]
$v$: speed [m/s]
$L$: characteristic length [m]
$\eta$: dynamic viscosity [Pa*s]
$F$: force [N]
$m$: mass [kg]
$x$: position [m]
$t$: time [s]
$\eta$: dynamic viscosity [Pa*s]
$a$: radius of particle [m]
$v$: speed [m/s]
Stokes' law states that the force of viscosity for a
spherical object moving through a viscous fluid is given
by $F=6\pi\eta av$ with $a$ the radius of the sphere. Importantly,
this force is proportional to the velocity $v$.
$v$: speed [m/s]
$v_0$: initial speed [m/s]
$\eta$: dynamic viscosity [Pa*s]
$a$: radius of cell [m]
$m$: mass [kg]
$t$: time [s]
$t_0$: start time [s]
To calculate the mass of a bacterium, we assume that the
bacterium is spherical with radius $a$ and the density
is twice that of water. Therefore, $m=4/3 \cdot \pi \cdot a^3 \cdot (2 \rho)$.
$\mu$: bacterial diffusion constant [µm^2/h]
$\vec{W}$: Wiener process [h^(1/2)]
$\Delta t$: timestep [h]
$\vec{N}$: normal distribution [-]
$\vec{N}$ indicates a vector of which the components
are independent random variables that are normally
distributed.
$\mu$: bacterial diffusion constant [µm^2/h]
$\Delta t$: timestep [h]
$\vec{N}$: normal distribution [-]
$X_1$: coordinate at next timestep [µm]
$x_1$: coordinate at current timestep [µm]
$\mu$: bacterial diffusion constant [µm^2/h]
$\Delta t$: timestep [h]
$N$: normal distribution [-]
$R$: intercellular distance at next timestep [µm]
$X_1$: coordinate at next timestep [µm]
$X_1$: coordinate at next timestep [µm]
$x_1$: coordinate at current timestep [µm]
$\mu$: bacterial diffusion constant [µm^2/h]
$\Delta t$: timestep [h]
$N$: normal distribution [-]
$\lambda$: noncentrality parameter [-]
$x_1$: coordinate at current timestep [µm]
$\mu$: bacterial diffusion constant [µm^2/h]
$\Delta t$: timestep [h]
$r_0$: intercellular distance at current timestep [µm]
$\bar{R}$: normalized intercellular distance at next timestep [µm]
$X_1$: coordinate at current timestep [µm]
$\mu$: bacterial diffusion constant [µm^2/h]
$\Delta t$: timestep [h]
$\chi^2$: chi-squared distribution [-]
$r_0$: intercellular distance at current timestep [µm]
$\lambda$: noncentrality parameter [-]
$\chi^2(k,\lambda)$ is the noncentral chi-squared distribution
with $k$ degrees of freedom and noncentrality parameter $\lambda$.
$F$: cumulative probability function, returns the probability that $\bar{R}^2$
is smaller than a given $\bar{R^*}^2$ [-]
$\bar{R^*}$: given scaled and squared intercellular distance for which
the cumulative probability function is calculated [µm]
$P$: probability [-]
$\bar{R}$: normalized intercellular distance at next timestep [µm]
Note that $F$ is defined with respect to the random variable $\bar{R}^2$ whose
distribution is given in (23). Therefore, $F$ depends on the
same parameters as $\chi^2$, which are the number of degrees of freedom $k$ and
noncentrality parameter $\lambda = \frac{r^{2}_{0}}{4 \cdot \mu \cdot \Delta t}$.
$F^{-1}$: inverse cumulative probability function, returns $\bar{R^*}^2$ that
$\bar{R}^2$ will not exceed with given probability $P$ [-]
$P^-$: desired probability of intercellular distance at next timestep exceeding
the cutoff distance, $1-P^-$ gives the probability that it will not be greater [-]
$\bar{r}_{cutoff}$: at distances greater than this cutoff distance the cell
no longer interact [µm]
$\mu$: bacterial diffusion constant [µm^2/h]
$\Delta t$: timestep [h]
The Hybrid Model
Introduction
The hybrid model represents an intermediate level of detail in between the
colony level model and the internal model. Bacteria are treated as individual agents
that behave according to Keller-Segel type stochastic differential
equations, while chemical species are modeled using partial differential
equations. These different models are implemented and coupled within a single
hybrid modeling framework.
Partial Differential Equations
Spatial reaction-diffusion models that rely on
Partial Differential Equations (PDEs) are based on the assumption that the
collective behavior of individual entities, such as molecules or bacteria, can
be abstracted to the behavior of a continuous field that represents the density
of those entities. The brownian motion of molecules, for instance, is the result
of inherently stochastic processes that take place at the individual molecule
level, but is modeled at the density level by Fick’s laws of diffusion. These
PDE-based models provide a robust method to predict the evolution of large-scale
systems, but are only valid when the spatiotemporal scale is sufficiently large
to neglect small-scale stochastic fluctuations and physical granularity.
Moreover, such a continuous field approximation can only be made if the behavior
of the individual entities is well described.
Agent-Based Models
Agent-based models on the other hand explicitly treat the entities as individual
“agents” that behave according to a set of “agent rules”. An agent is an object
that acts independently from other agents and is influenced only by its local
environment. The goal in agent-based models is to study the emergent
systems-level properties of a collection of individual agents that follow
relatively simple rules. In smoothed particle hydrodynamics for example, fluids
are simulated by calculating the trajectory of each individual fluid particle at
every timestep. Fluid properties such as the momentum at a certain point can
then be sampled by taking a weighted sum of the momenta of the surrounding fluid
particles. A large advantage of agent-based models is that the agent rules are
arbitrarily complex and thus they allow us to model systems that do not
correspond to any existing or easily derivable PDE model. However, because every
agent is stored in memory and needs to be processed individually, simulating
agent-based models can be computationally intensive.
Hybrid Modeling Framework
In our system, there are both bacteria and chemical species that spread out
and interact on a petri dish to form patterns. On the one hand, the bacteria are
rather complex entities that move along chemical gradients and interact with one
another. Therefore they are ideally modeled using an agent-based model. On the
other hand, the diffusion and dynamics of the chemicals leucine and AHL are
easily described by well-established PDEs. To make use of the advantages of each
modeling approach, we decided to combine these two different types of modeling
in a hybrid modeling framework. In this framework we modeled the bacteria as
agents, while the chemical species were modeled using PDEs. There were two
challenges to our hybrid approach, namely coupling the models and matching them.
Coupling refers to the transfer of information between the models and matching
refers to dealing with different spatial and temporal scales to achieve
accurate, yet computationally tractable simulations.
In the following paragraphs we first introduce our hybrid model and its
coupling. Once the basic framework is established, the agent-based module and
PDE module are discussed in more depth and the issue of matching is highlighted.
We also expand on important aspects of the model and its implementation such as
boundary conditions and choice of timesteps. Then the results for the 1-D model
and 2-D model simulations are shown and summarized. Finally, the incorporation
of the internal model into the hybrid model is discussed and a proof of concept
is demonstrated.
Model Description
System
The main protagonists in our pattern-forming system are cell types A and B,
AHL and leucine. Cells A produce AHL as well as leucine. They are unaffected by
leucine, while cells B are repelled by leucine. AHL modulates the motility of
both cell types A and B, but in opposite ways. High concentrations of AHL will
render cell type A unable to swim but will activate cell type B’s motility.
Conversely, low concentrations of AHL causes swimming of cell type A and
incessant tumbling (thus immobility) of cell type B. Lastly, cells A express the
adhesin membrane protein, which causes them to stick to each other. Simply said,
our system should produce spots of immobile, sticky groups of A type cells,
surrounded by rings of B type cells. Any cell that finds itself outside of the
region that it should be in, is able to swim to their correct place and becomes
immobile there. More details can be found in the
research section.
Partial Differential Equations
As discussed in the previous paragraph, our hybrid model incorporates chemical
species using PDEs. In our system these are AHL and leucine. The diffusion of
AHL and leucine can be described by the heat equation (1).
By using (1)
we assume that the diffusion speed is isotropic, i.e. the same in all spatial
directions. This also explains why it is called the heat equation, since heat
diffuses equally fast in all directions. A detailed explanation of the heat
equation can be found in box 1. The second factor that needs be taken into
account is the production of AHL and leucine by type A bacteria. In principle,
AHL and leucine production is dependent on the dynamically-evolving internal
states of all cells of type A. However, for our hybrid model we ignored the
inner life of all bacteria and instead assumed that AHL and leucine production
is directly proportional to the density of A type cells (2).
In the last
paragraph we will reconsider this assumption and assign each cell an internal
model. Finally, AHL and leucine are organic molecules and therefore they will
degrade over time.
We assume first-order kinetics meaning
that the rate at which AHL and similarly leucine disappear is proportional to
their respective concentrations (3a and 3b) assuming neutral pH
[6].
Note that these equations have exactly the same form as the equations for AHL and leucine
in the colony level model. The crucial difference however lies in the
calculation of the density of cells of type A. In contrast to the colony level
model, in this model the cell density is
not calculated explicitly with a PDE and is therefore
not trivially known. Therefore a method to extract a density field from a
spatial distribution of agents is necessary. This is addressed in the
subparagraph below on coupling.
Box 1: Heat Equation
Heat Equation
The left-hand side of the heat equation (1) represents the rate of
accumulation of a chemical, while the right-hand side is proportional
to the Laplacian
of its concentration field, which is a second-order differential operator.
This equation can be easily understood by
considering a one-dimensional concentration profile, as shown in Figure 1: if the concentration
can be approximated as a convex parabolic function, the second derivative
is positive and therefore the rate of accumulation is positive (i.e. more
accumulation). If on the other hand the concentration resembles a concave
parabolic function, the second derivative is negative and the rate of
accumulation as well (i.e. depletion). A special case occurs when the
concentration profile takes on a linear form. Everything that moves into
the point goes out at the other side and as a result there is no net accumulation
over time.
Figure 1
Illustration of the heat equation. Click to enlarge.
In the videoplayer below we demonstrate how the heat equation smoothes out
an initially heterogeneous concentration profile. When only diffusion is
acting on the system, it will always evolve to a uniformly flat
concentration profile, regardless of the initial conditions.
Agents
To model bacteria movement on the other hand, we used an agent-based model that
explicitly stored individual bacteria as agents. Spatial coordinates are
associated with each agent, specifying their location. After solving the
equation of motion for all agents based on their environment, these coordinates
are updated at every timestep. In principle, Newton’s second law of motion has
to be solved for all bacteria. However, since bacteria live in a low Reynolds
(high friction) environment, the inertia of the bacteria can be neglected. This
is because an applied force will immediately be balanced out by an opposing
frictional force, with no noticeable acceleration or deceleration phase taking
place.
This eliminates the inertial term and simplifies Newton’s second law to
(5).
Basically, the velocity can be calculated as the sum of all applied
forces times divided by a frictional coefficient.
For more info about the Reynolds number and
“life at low Reynolds number”, we refer to box 2. In the following
paragraphs we will investigate the different forces acting on the bacteria
and ultimately superimpose them to obtain the final equation of motion.
Box 2: Life at Low Reynolds Number
Life at Low Reynolds Number
The Reynolds number in fluid mechanics is a dimensionless number that characterizes
different flow regimes. It is most commonly used to determine whether laminar
or turbulent flow will take place in a hydrodynamic system (Figure 2).
Figure 2
Flow regimes for different Reynolds numbers. Click to enlarge.
In general however, it quantifies the ratio of inertial forces to viscous forces
and is defined as (B2.1).
$$
Re=\frac{\text{inertial forces}}{\text{viscous forces}}
=\frac{\rho v^2 L^2}{\eta v L}
=\frac{\rho v L}{\eta} \;\;\; \text{(B2.1)}
$$
For example, if the Reynolds number is high, the inertia of fluids in motion dominate and
turbulent flow will occur. When it is low however, the viscous forces dampen
the kinetic energy of fluid particles and stabilize the flow profile, ultimately
achieving a regular, laminar flow.
The reason we mention the Reynolds number however is not to study the flow of fluids, but
to characterize the behavior of swimming objects inside of a stationary fluid.
When applied to objects in a fluid, the Reynolds number tells us whether their inertia
can be neglected or not. For bacteria, the characteristic length $L$ is on the order
of micrometers, which is quite small. Therefore the Reynolds number is always
very small and hence viscous forces dominate the motion of bacteria. This is what allows
us to eliminate the inertial term in Newton's second law and greatly simplify the
equation of motion.
To justify this simplification we will go through a numerical example.
Take a bacterium of size $L = 2 a = 1 \cdot \mu m$, swimming through water with a speed of
$v=20 \cdot \mu m/s$. Water has a density of $\rho = 1 \; 000 \cdot kg/m^3$
and a viscosity of $\eta = 0.001 \cdot Pa/s$.
Figure 3
Sketch of swimming bacterium. Click to enlarge.
Putting these values in (B2.1) yields
a Reynolds number of $Re = 2 \cdot 10^{-5} << 1$. Clearly we are in an extremely
low Reynolds number regime. To show what this really means, suppose the bacterium
stops propelling itself. How long will it continue to move relying only on its inertia?
Assuming Stokes' law, we obtain (B2.2) as the equation of motion.
The characteristic time constant is $\tau = 6\pi\eta a/m \approx 0.1 \cdot \mu s$,
from which we calculate that the total distance traversed is $\Delta x < 2 \cdot pm$.
This coasting distance is 6 orders of magnitude smaller than its size. Moreover,
the time spent coasting is extremely short. Thus, once the
bacterium stops propelling itself, it is safe to assume that whatever kinetic energy
it had is immediately absorbed by hydrodynamic friction, instantly halting the bacterium.
Therefore, we can neglect the inertial term in Newton's second law (5).
Stochastic Differential Equation
When we try to calculate the physical “chemotactic force”, propelling bacteria towards
chemoattractants or away from chemorepellents, we find that it is not easily
measured nor derived. Therefore, as a workaround
we base the equation of motion
on (6), a stochastic differential equation (SDE) that describes the motion
of a single particle in a N-particle system that is governed by a Keller-Segel
type PDE in the limit of $N \rightarrow \infty$. This Keller-Segel type PDE (7)
describes the evolution of a bacteria density field $n$ moving
towards some chemoattractant $S$.
Put differently, when infinitely many particles obey (6), they will exhibit
Keller-Segel type spatial dynamics as in (7). In a sense,
we’re using a “reverse-engineered” particle equation that corresponds to the
Keller-Segel field equation. A detailed theoretical treatment of (6) is
outside the scope of this model description because it contains a stochastic
variable. The traditional rules of calculus do not apply anymore for stochastic
differential equations and a different mathematical theory called Itô calculus
is required. It is sufficient to say that the second term containing
$dW$ accounts
for Brownian motion in the form of random noise added to the displacement of the
agents, causing them to diffuse, and that it is governed by the diffusion
coefficient $\mu$.
The first term in (6) on the other hand is easily understood
as an advective or drift term (net motion) dependent on S, pushing the agents
along a positive gradient (for negative chemotaxis the sign is reversed). The
chemotactic drift hence points towards an increasing concentration of the
chemoattractant. The advective properties are governed by the chemotactic
sensitivity function $\chi (S)$. For our model we used a function of the
form (8).
The first important thing to note is that we assume
$\chi (S)$ to be proportional to $1/S$. This is because Keller and Segel proved that
their corresponding PDE model only yields travelling wave solutions if $\chi (S)$
contains such a singularity at some critical concentration $S_{crit}$,
and multiplying by
$1/S$ is the simplest way to introduce a singularity at $S_{crit} = 0$. Secondly, the
proportionality constant is composed of two factors, namely the bacterial
diffusion coefficient $\mu$ and chemotactic sensitivity constant $\kappa$.
This is done for
two reasons. Firstly, when $\mu$ is lowered, both chemotactic and random motion is
reduced, which emulates the state of inactivated motility due to high or low
concentrations of AHL. Secondly, defining a separate chemotactic sensitivity
constant allows us to examine the effect of the relative strength of chemotaxis
versus random motion on pattern formation. Conceptually, this term can be
viewed as a sort
of chemotactic force, defined with regards to a mobility coefficient $\mu$
instead of a frictional coeffient, which are the inverse of each other.
Cell-Cell Interactions
In addition to chemotaxis and diffusion, cell-cell interactions play an
important role in pattern formation and also need to be modeled. Bacteria have
finite size and therefore multiple bacteria cannot occupy the same space.
Moreover, an important mechanism in our system is the aggregation of cells A due
to the sticky adhesin protein membrane. To take these mechanisms into account we
modeled two types of cell-cell interactions: the purely repulsive interaction of
cell B with another cell B and with cell A, and the attractive-repulsive
interaction of cell A with another cell A. The interaction between two cells is
usually expressed by a potential energy curve defined over the distance between
the centers of mass of the two cells (Figure 2).
Figure 2
Cell-cell interaction potential. Click to enlarge.
Note that the potential energy
remains constant after a certain distance, which means that the cells stop
interacting. Also, as two cells move closer together, they hit a wall where the
potential energy curve abruptly goes to infinity. The reason for this is that
two cells cannot occupy the same space and therefore smaller intercellular
distances are not allowed. Implementing this theoretical potential is however
not possible because the displacement of bacteria is a stochastic process
and the bacteria could randomly jump beyond
the potential wall, where the force is ill defined. Therefore, we’ve decided
to define a piecewise quadratic potential (9),
which results in a piecewise
linear force that resembles Hooke’s law, but with three different “spring
constants” acting in different intervals of intercellular distances (10),
as illustrated in Figure 3. The force is defined with respect to
the unit vector pointing towards the other cell, meaning that a positive
force corresponds to an attractive force and vice-versa.
Figure 3
Cell-cell interaction potential and force curves. Click to enlarge.
Between A type cells, there is a region of attraction
$(R_0 \leq r_{ij} < R_{cutoff})$,
where the force points towards the other cell,
hence moving them closer together.
In the repulsive domain $(r_{ij} < R_0)$,
two regions were defined, emulating lower
repulsive forces $(\frac{R_0}{2} \leq r_{ij} < R_0)$
and higher repulsive forces due to a higher spring
constant when the cells are even closer $(r_{ij} < \frac{R_0}{2})$.
For the purely repulsive interaction
scheme there is no attraction and therefore
the spring constant for $R_0 \leq r_{ij}$ is zero.
More details about the implementation of the
cell-cell interaction scheme, more specifically
regarding the nearest-neighbor search algorithm,
can be found in the paragraph on the agent-based
module below.
Equation of Motion
Now we are ready to construct the equation of motion for cell type A and B as a
superposition of an adapted Keller-Segel SDE (6)
and cell-cell interaction forces (10),
yielding (11).
Bacteria of type A are not attracted nor repelled by leucine,
so the chemotactic term falls away. All cell-cell forces are summed up to find a
net force, taking into account the two different potentials due to the different
interaction types. As discussed before, this net force times a constant yields
the velocity due to that force, which is then multiplied by $dt$ to obtain the
displacement. For B type cells, the chemotactic term models the repulsive
chemotaxis away from leucine. The chemotactic sensitivity function (12) has a
negative sign signifying that B type cells are repelled by leucine. The cell-cell
interaction term in this case is simpler because B type cells only interact
repulsively.
Note that the diffusion coefficient of cell types A and B (13) switches
based on the local concentration of AHL relative to a threshold AHL value, which
simulates the dependency of cellular motility on AHL.
For B type cells the cellular motility depends explicitly on AHL due to the
synthetic genetic circuit we have built into them. On the other hand,
in our model the motility of A type cells should not depend on AHL directly,
but since high concentrations of AHL are caused by high densities of
aggregated A type cells, the bacteria typically will not be motile
in high AHL
concentrations because they stick to neighboring cells.
The agent-based module is
now fully defined but one crucial issue was skipped: AHL and leucine
concentrations are calculated using PDEs and are therefore only known at grid
points. Agents on the other hand reside in the space between grid points and
require local concentrations as inputs to calculate their next step. This
problem is part of the coupling aspect in our hybrid modeling framework and is
discussed below.
Coupling
At this point, both the PDE module and agent-based module have been established,
but the issue remains that the individual modules take inputs and return outputs which are not
directly compatible. In the framework of PDEs, entities are described by
density/concentration fields.
In the agent-based module however, entities are represented by discrete
objects with exact spatial locations. The key challenge posed by the hybrid model
is to integrate these different approaches by bidirectionally converting and
exchanging information between the modules. Coupling the modules is the essential component
that makes the hybrid model hybrid. It allows us to leverage the advantages
of both modeling approaches, while circumventing their drawbacks. But although
hybrid modeling opens up many new avenues for novel modeling methods,
it comes with its own diverse set of issues and peculiarities that need to be
addressed before it can be successfully applied.
In the following paragraphs the basic scheme for coupling
the PDE and agent-based modules in our model is introduced, after which the theoretical
treatment of our hybrid model is complete. The difficulties that arise from linking the
modules together are discussed further below in the subsection on matching.
Agent-Based to PDE
As described above, the agents’ effect in the PDE is modeled as a source term
that is proportional to the agent density. This approach is essentially the same
approach taken in the colony level model for the bacterial production of AHL and
leucine. However, in the colony level model the bacteria density is explicitly
calculated at the grid points, while the agent-based model essentially considers
a set of points scattered in space.
A simple first-order approach would be to determine
the closest grid point to any agent and simply increment a counter belonging to
that grid point. This results in a histogram, which can be used directly to
represent the agent density. However, the resulting density is a blocky,
nonsmooth function which poorly represents the underlying agent distribution.
The effect of a single agent is artificially confined to a single grid point,
while in reality an agent’s influence could reach much further than a single
grid point. The shape of a histogram is also very dependent on the bin size,
which in this case corresponds to the grid spacing so it cannot be independently
tuned. To decouple grid spacing and agent density and achieve a smoother density
function, we made use of a more sophisticated technique called kernel density
estimation (KDE).
KDE is used in statistics to estimate the probability density of a set discrete
data derived from a random process. The basic idea consists of defining a
kernel function that represents the density of a single data point, then
centering kernel functions on every data point and summing them all up to
achieve a smooth overall density function, as demonstrated in the Figure 4.
Figure 4
Kernel density estimation. Click to enlarge.
This
kernel function can be anything as long is it continuous, symmetric and
integrates to 1, since it represents one data point or one agent in our case.
Some of the most common kernel functions include Gaussian kernels, triangular
kernels and Epanechnikov kernels (14), which are shown in Figure 5.
Figure 5
Gaussian, triangular and Epanechnikov kernel functions.
Click to enlarge.
In practice, scaled versions of standard kernel functions are used,
which are of the form (15).
$$
K_h(x)=\frac{1}{h} \cdot K \bigg( \frac{x}{h} \bigg) \;\;\; \text{(15)}
$$
Importantly, the scaled functions inherit the kernel function
properties, but are either broader or narrower. The degree to which the shape of
a kernel function is stretched or squeezed depends on the scaling factor h,
which is why it is called the bandwidth, see Figure 6.
Figure 6
Epanechnikov kernels using different bandwidths. Click to enlarge.
This parameter gives us the
freedom to define how far the influence of an agent reaches and how smooth the
resulting density function looks like. An example is given in Figure 7, where
100 agents were sampled from a normal distribution. The red dotted line
indicates the true underlying distribution, while the blue line is the
kernel density estimation calculated using different bandwidths. The left graph
shows a undersmoothed (small bandwidth) KDE, which oscillates wildly.
The right graph on the other hand is oversmoothed (large bandwidth),
leading to a misleadingly wide KDE.
Figure 7
Kernel density estimation using different bandwidths. Click to enlarge.
In summary, using kernel density estimation we can express the agent density
in the form of (16), providing the appropriate input for the PDE model.
The extension to higher dimensions is called multivariate kernel density
estimation and is rather non-trivial since the bandwidth is then defined
as a matrix instead of a scalar, which allows for many more variations
on the same basis kernel function. A detailed analysis of multivariate
kernel density estimation will not be given here. It suffices to
say that for our 2-D hybrid model we used the simplest
possible bandwidth matrix, which is the identity matrix times a constant.
PDE to Agent-Based
The final component of our hybrid model is the mapping
of the PDE model to the agent-based model.
The latter model works with discrete objects that
have continuous coordinates, meaning that they can be located
at any point of the domain.
As we have seen in (11), the agents need the local concentration
of AHL and leucine, as well as the gradient of leucine in
order to update their positions. In the PDE model however,
the domain is discretized into a grid and concentrations
are only defined at grid points. Therefore, in order to
transfer information from the PDE model to the agent-based
model we need to translate these grid values into values
for any given position within the domain. We achieved this
for the 1-D hybrid model by taking the two nearest grid points
to any agent and employing linear interpolation to
derive an approximate local field value. Similarly, for the 2-D
hybrid model we took the four nearest grid points
and employed bilinear interpolation, as explained in box 3.
With the end result (B3.3), the concentration or gradient of AHL
or leucine can be derived at any point of the domain. Therefore,
the agent-based module has access to the information that it needs
from the PDE module.
Box 3: Bilinear Interpolation
Bilinear Interpolation
Interpolation is used to obtain a continuous function
when a finite amount of data points are known. Piecewise
linear interpolation is the most simple scheme to interpolate
1-dimensional data.
Figure 8
Illustration of linear interpolation. Click to enlarge.
The technique involves defining
linear functions connecting consecutive data points to fill in
the gaps, using (B3.1).
Bilinear interpolation is the generalization
of this interpolation technique to two dimensions,
illustrated using Figure 9.
Figure 9
Illustration of bilinear interpolation. Click to enlarge.
Let’s assume we want to interpolate
a two-dimensional function within a rectangular domain where
the function values are only known on the four adjacent grid
points $Q$. First we linearly interpolate in the x-direction on
the lines connecting $Q_{i,j}$ and $Q_{i+1,j}$,
$Q_{i,j+1}$ and $Q_{i+1,j+1}$,
which yields (B3.2).
We then obtain interpolated
values for all positions on the bottom and top edge. The key
idea then is to take these values and interpolate in the y-direction
for any given x-position, yielding (B3.3).
This then gives
interpolated values for any point within the domain. Note that
the resulting function is not linear, but quadratic since it contains the term
$x \cdot y$.
Synthesis
The diffusion, production and degradation of AHL and leucine are
described by the PDEs (4). At the same time, the random movement,
chemotaxis and intercellular
interactions of cell types A and B are captured by
the stochastic equation of motion (11). Translating the spatial distribution of
cells type A into a density field to feed into the PDE module can be
accomplished by using kernel density estimation (16). Finally, the agents
can request concentrations and gradients at arbitrary positions from the
PDE module using
bilinear interpolation (B3.3). Taken together, these equations describe the
individual modules, as well as the two-way information transfer in between them,
and thus they fully define our hybrid model. A graphical summary is shown in
Figure 10.
Figure 10
Graphical summary of hybrid model. Click to enlarge.
Implementation
Agent-Based Module
In this paragraph, we will discuss the implementation of
the agent-based module. First, the problem of computing
pairwise interactions is discussed and a nearest-neighbor
algorithm is introduced to reduce the computation time.
Then the choice of a timestep is considered especially
with regards to the cell-cell interaction scheme. Finally,
the boundary conditions for agents are defined and explained.
Nearest-Neighbor Algorithm
The cell-cell interactions have already been fully described in the paragraphs above. However, solving the equation of motion of an agent in its current form requires the computation of the force due to every other agent. If we take N to be the number of agents, that means $N*(N-1)$ amount of force computations are needed in total and therefore the computation time grows as $O(N^2)$. To put this in perspective, if we simulate a thousand agents, the amount of interactions adds up to around one million. This puts a heavy computational load on the model which can be reduced greatly by using a smarter algorithm. The crucial observation here is that the force goes to zero when the distance between two cells is larger than $2\cdot r_{cutoff}$. An agent therefore only interacts with its nearest neighbors, meaning that the naive implementation wastes a lot of time computing interactions which have no effect anyways. Moreover, as the simulation progresses, an agent’s neighbors do not vary greatly from one timestep to another. A more efficient approach then consists of periodically searching for the nearest neighbors within a fixed radius, storing them in a list for every agent and updating their coordinates for all intermediate timesteps.
Since the agents are programmed to repel each other when they approach each other too closely, they will evolve to a rather uniform distribution. The most appropriate fixed-radius nearest neighbor search algorithm in that case is the cell technique. In this algorithm, the agents are structured by placing a rectangular grid of cells over the domain and assigning every agent to a cell (fig9). Determining which cell an agent belongs to is easily done by rounding down the x and y-coordinates to the nearest multiple of the grid spacing. If the grid spacing is chosen so that interactions do not reach further than adjacent cells, every agent only needs to look for neighbors in 9 cells (its own cells plus 8 adjacent cells) instead of the entire domain.
Choice of Timestep
In contrast to PDE models, there is no danger of instability
due to a large timestep in agent-based models. However, since
the time step determines how far a cell can “jump” away from
its previous position, we need to ensure that it is not so
large as to negate the effect of intercellular interactions.
To understand this, we first discretize the Brownian motion term
in the agent’s equation of motion (17).
The details of this discretization scheme goes beyond the scope of this
text and will not be given here.
It can be proven that
this term follows a distribution as (18).
Now assume that there
are two A type cells who are exactly $2 \cdot r_0$ apart.
If the timestep
is too large, they might “jump” out of each other’s influence
at the next iteration. This situation is highly unrealistic
because the cells actually should feel each other’s pull while
they randomly drift away from each other. If the timestep is
too large, there’s a high probability that the cells detach
without experiencing the potential at all and therefore the
probability of detachment is artificially inflated.
Of course, we can never completely eliminate the possibility
of a sudden detachment because the random is sampled by a
normal distribution, but we can define a desired probability
of sudden detachment $P^-$. Then, for a given $r_0$,
$r_{cutoff}$ and $\mu$,
we can determine how small the timestep has to be for the
sudden detachment to happen only with probability $P^-$. First,
we assume two A type cells with coordinates $(x_1,y_1)$ and
$(x_2,y_2)$ that are at a distance $2 \cdot r_0 = R_0$
from each other.
We then define the coordinates at the next timestep as
stochastic variables $X_1$, $Y_1$, $X_2$ and $Y_2$, which are distributed
as (19).
We are not interested in the future coordinates
as such, but rather in the future intercellular distance,
which is composed of these future coordinates and therefore
also is a stochastic variable. For ease of calculation, we
work with the square of this variable, which we name as R²
and define as (20).
It is easy to see that $X_1-X_2$ and
$Y_1-Y_2$ are both noncentral (mean is not equal to zero)
normally distributed variables as well. Therefore $R^2$ obeys
a noncentral chi-squared distribution with 2 degrees of freedom.
Before proceeding further, we have to make our $R^2$ conform
to a standard noncentral chi-squared distribution. First
we normalize $X_1-X_2$ and $Y_1-Y_2$. We observe that $X_1-X_2$
and
$Y_1-Y_2$ are distributed as (21).
Secondly, because the
means of the $X_1-X_2$ and $Y_1-Y_2$ are not equal to zero, we
need to account for this by calculating the noncentrality
parameter $\lambda$. This is equal to the sum of squares of
the means of the normalized $X_1-X_2$ and $Y_1-Y_2$ variables,
which of course equals (22), corresponding to a scaled
version of the original intercellular distance.
To recap,
the meaning of $\bar{R}^2$ is the scaled squared intercellular distance
after updating once with timestep $\Delta t$ starting from
an initial distance of $r_0$. Then we define $F$ as the cumulative
distribution function for the random variable $\bar{R}^2$ as (24).
To then finally calculate
appropriate timestep for a given $r_0$, $r_{cutoff}$, $\mu$
and desired $P^-$
we take
the inverse cumulative distribution function and equate it
to the intercellular cutoff distance, after appropriate scaling
and squaring (25).
Note that both the left hand side and right hand side both
depend on the $\Delta t$ in a nonlinear way, therefore
a nonlinear solver is required to solve this equation.
As stated previously, it is not possible to completely avoid
that the cells will jump out of each other's influence in a single
timestep. However, equation (25) allows us to calculate the timestep
$\Delta t$ that will result in cells detaching in a single timestep with a
probability $P^-$ that we can specify.
Partial Differential Equations Module
Introductory text about the Partial Differential Equations Module
Computational Scheme
As mentioned earlier the concentrations of AHL and Leucine
are modeled using partial differential equations.
In the colony level model these equations are solved
explicitly. Explicit schemes do not require a lot of
work per time step, but unfortunately are not
unconditionally stable. In two dimensions the grid ratios
$dt/dx^2$ and $dt/dy^2$ can not exceed $dt/dx^2 + dt/dy^2
\leq \frac{1}{2}$ for the solver to be stable.
When computing the solution of the hybrid model this
requrement forces us to spend a lot of CPU time solving
partial differential equations that could be better spent
simulation the agents. Therefore an implicit ADI
Alternating direction implicit scheme has been implemented.
ADI-schemes are unconditionally stable, which
allows it to take large time steps with the PDE solver.
We used the following scheme:
$$ (1 - \frac{1}{2} \mu_x \delta_x^2) U^{n+\frac{1}{2}}
+ \frac{1}{4}kU^{n+\frac{1}{2}}
= (1 + \frac{1}{2} \mu_y \delta_y^2) U^n
- \frac{1}{4}kU^n + \frac{\alpha}{2} \rho_A $$
$$ (1 - \frac{1}{2} \mu_y \delta_y^2) U^{n+1}
+ \frac{1}{4}kU^{n+1} =
(1 + \frac{1}{2} \mu_x \delta_x^2)U^{n+\frac{1}{2}}
- \frac{1}{4}kU^{n+\frac{1}{2}} + \frac{\alpha}{2} \rho_A $$
In the equations above $\mu$ denotes grid ratios and $\delta^2$
central differences. The production and
degradation terms have been incorporated at every time
level with a factor of $\frac{1}{4}$.
The image below shows the computational molecule of the ADI
scheme we chose to implement:
Figure 4
ADI-Molecule. Click to enlarge.
Boundary Conditions
Text about boundary conditions.
Matching
Introductory text about matching
Decoupling of Timesteps
In order to benefit from the implicit PDE-solver described above the agent's time steps are chosen smaller
then the time steps of the PDE solver. However type A cells produce molecules continuously as they move
trough space. Therefore we record their positions and average over the kernel functions centered at
past positions since the last PDE evaluation. That way we avoid blurring the results of the PDE solver
too much, when the time step of the agents is reduced.
Last but not least we want to point out the relationship between kernel bandwidth and the grid on which the
PDE is solved. The larger the kernel bandwidth $h$ is chosen the coarser the PDE grid can be without loosing
cells between grid points. If the the diameter of the kernel functions is smaller then the distance between
PDE grid points it can happen, that the kernel does not overlap with any PDE grid points. In this case a cells
contribution to the molecule concentrations at the next time step could be lost. When the PDE grid is
widened to save computing time it is thus necessary to increase the Kernel bandwidth as well. However this
increase can lead to a situation where the Kernel function covers significantly more space then the actual
diameter of a bacterium. In these cases the Kernel function can be interpreted as a probability function for
a cells position. However we avoided too large bandwiths and PDE grids by running our 2-D simulations at the
Flemish Supercomputer Center (VSC).
Figure 6
Logo of the Flemish Supercomputer Center. Click to enlarge.
Bandwidth Tuning
Text about bandwidth tuning.
1-D Hybrid Model
The video box above shows one dimensional simulation results for the hybrid model. A constant speed and random step
simulation has been computed. We observe that the bacteria form a traveling wave in both cases, which is essential
for pattern formation. These results are also similar to what we get from the continuous model, which confirms our
results.
2-D Hybrid Model
The videos above show simulation videos computed at the Flemish supercomputing center, for three different
initial conditions similar to the ones we used for the colony level model. The first and second condition start
from 9 mixed or 5 colonies of both cell types, arranged in a block or star shape. These first two gradually
separate in a manner similar to what we would we also saw in the colony level model. The result for random
initial data is fundamentally different. As the agent based approach allows for better implementation
of adhesion large cell type A bands form. The AHL and Leucine produced by the type A bacteria causes the
B type cells to move away leading to a pattern which we could not produce using PDEs alone, this
beautifully illustrates the added value of hybrid modeling.
Incorporation of internal model
Up until now, we have largely ignored the inner life of the bacteria. This inner life consists of transcriptional networks and protein kinetics. Instead we assumed that AHL and leucine production is directly proportional to the density of type A cells. This only works in theory, since bacteria will be affected by their surroundings and the way their dynamics react to it. For example bacteria surrounded by a large concentration of AHL, will have more CheZ and will react more on the presence of Leucine. Also bacteria have different histories and will have different levels of transcription factors and different levels of proteins in their plasma. The proteins are not directly degraded and will still be present in the cytoplasm of the bacteria long after the network has been deactivated. From this, it is clear that 2 bacteria, although surrounded by the same AHL and leucine concentrations, can show different behavior and reaction kinetics.
This results in a heterogeneity of the bacterial population that has not yet been accounted for. To make up for this anomaly, we decided to add an internal model to every agent. This way we will get more realistic simulations. Every agent will get their own levels of CheZ, LuxR, LuxI and so on and will have individual reactions on their surroundings. We hope that this way we can get closer to the behavior of real bacteria.
Benjamin Franz and Radek Erban.
Hybrid modelling of individual movement and collective behaviour.
Lecture Notes in Mathematics, 2071:129-157, 2013.
[ .pdf ]
Zaiyi Guo, Peter M A Sloot, and Joc Cing Tay.
A hybrid agent-based approach for modeling microbiological systems.
Journal of Theoretical Biology, 255(2):163-175, 2008.
[ DOI ]
E F Keller and L A Segel.
Traveling bands of chemotactic bacteria: a theoretical analysis.
Journal of theoretical biology, 30(2):235-248, 1971.
[ DOI ]
A. L. Schaefer, B. L. Hanzelka, M. R. Parsek, and E. P. Greenberg.
Detection, purification, and structural elucidation of the
acylhomoserine lactone inducer of Vibrio fischeri luminescence and other
related molecules.
Bioluminescence and Chemiluminescence, Pt C, 305:288-301,
2000.