MBI Videos

2012 MBI-CAMBAM-NIMBioS Summer Graduate Program: Stochastics Applied to Biological Systems

  • video photo
    Edward Allen
    Properties of the Wiener process are reviewed and stochastic integration is explained. Stochastic differential equations are introduced and some of their properties
    are described. Equivalence of SDE systems is explained. Commonly used numerical
    procedures are discussed for computationally solving systems of stochastic differential
    equations. A procedure is described for deriving Itˆo stochastic differential equations
    from associated discrete stochastic models for randomly-varying problems in biology.
    The SDEs are derived from basic principles, i.e., from the changes in the system which
    occur in a small time interval. Several examples illustrate the procedure. In particular, stochastic differential equations are derived for predator-prey, competition, and
    epidemic problems.
  • video photo
    Scott McKinley
    Rapid recent progress in advanced microscopy has revealed that nano-particles
    immersed in biological
    uids exhibit rich and widely varied behaviors. In some
    cases, biology serves to enhance the mobility of small scale entities. Cargo-laden
    vesicles in axons undergo stark periods of forward and backward motion, inter-
    rupted by sudden pauses and periods of free di usion. Over large periods of time,
    the motion is e ectively that of a particle with steady drift accompanied a di u-
    sive spread greater than what can be explained by thermal
    uctuations alone. As
    another example, E. coli and other bacteria are known to respond to the local con-
    centration of nutrients in such a way that they can climb gradients toward optimal
    locations. Again, the e ective behavior is drift toward a desired" location, with
    enhanced di usivity.
    In other cases, biological entities are signi cantly slowed. Relatively large parti-
    cles di using in
    uids such as mucus, blood, bio lms or the cytoplasm of cells all
    experience hinderances due to interactions with the polymer networks that consti-
    tute small-scale biological environments. Researches repeatedly observe sublinear
    growth of the mean-squared displacement of particle paths. This signals to theo-
    reticians that the particles are not experiencing traditional Brownian motion. In-
    terestingly, many viruses are actually small enough to avoid this type of hinderance
    when moving through human mucus. However, the body's immune response in-
    cludes teams of still smaller antibodies that can immobilize virions by serving as an
    intermediary creating binding events between virions and the local mucin network.
    Underlying the mathematical description of all these phenomena is a modeling
    framework that employs stochastic di erential equations, hybrid switching di u-
    sions and stochastic integro-di erential equations. We will begin with the Langevin
    model for di usion. This is the physicist's view of Brownian motion, derived from
    Newton's Second Law. We will see how the traditional mathematical view of Brow-
    nian motion arises by taking a certain limit. The force-balance view permits a
    variety of generalizations that include particle-particle interactions, the in
    uence of
    external energy potentials, and viscoelastic force-memory e ects. We will use sto-
    chastic calculus to derive important statistics for the paths of such particles, develop
    simulation techniques, and encounter a number of unsolved theoretical problems.
  • video photo
    Scott McKinley
    Rapid recent progress in advanced microscopy has revealed that nano-particles
    immersed in biological
    uids exhibit rich and widely varied behaviors. In some
    cases, biology serves to enhance the mobility of small scale entities. Cargo-laden
    vesicles in axons undergo stark periods of forward and backward motion, inter-
    rupted by sudden pauses and periods of free di usion. Over large periods of time,
    the motion is e ectively that of a particle with steady drift accompanied a di u-
    sive spread greater than what can be explained by thermal
    uctuations alone. As
    another example, E. coli and other bacteria are known to respond to the local con-
    centration of nutrients in such a way that they can climb gradients toward optimal
    locations. Again, the e ective behavior is drift toward a desired" location, with
    enhanced di usivity.
    In other cases, biological entities are signi cantly slowed. Relatively large parti-
    cles di using in
    uids such as mucus, blood, bio lms or the cytoplasm of cells all
    experience hinderances due to interactions with the polymer networks that consti-
    tute small-scale biological environments. Researches repeatedly observe sublinear
    growth of the mean-squared displacement of particle paths. This signals to theo-
    reticians that the particles are not experiencing traditional Brownian motion. In-
    terestingly, many viruses are actually small enough to avoid this type of hinderance
    when moving through human mucus. However, the body's immune response in-
    cludes teams of still smaller antibodies that can immobilize virions by serving as an
    intermediary creating binding events between virions and the local mucin network.
    Underlying the mathematical description of all these phenomena is a modeling
    framework that employs stochastic di erential equations, hybrid switching di u-
    sions and stochastic integro-di erential equations. We will begin with the Langevin
    model for di usion. This is the physicist's view of Brownian motion, derived from
    Newton's Second Law. We will see how the traditional mathematical view of Brow-
    nian motion arises by taking a certain limit. The force-balance view permits a
    variety of generalizations that include particle-particle interactions, the in
    uence of
    external energy potentials, and viscoelastic force-memory e ects. We will use sto-
    chastic calculus to derive important statistics for the paths of such particles, develop
    simulation techniques, and encounter a number of unsolved theoretical problems.
  • video photo
    Steve Krone
    We will work through some of the basic ideas involved in modeling various types of interactions in spatial population biology using interacting particle systems (sometimes referred to as stochastic cellular automata). Some of the essential ingredients and behaviors come from simple models like the contact process and the voter model. These components can be combined and tweaked to obtain models with more biological detail, including epidemic behavior for host-pathogen systems, the spread of antibiotic resistance genes, etc. These models can be informative since real biological populations exhibit a high degree of spatial structure and this structure affects the interactions between individuals and species in ways that can dramatically alter dynamics compared to well-mixed systems. The computer exercises will allow students to alter some existing MATLAB code to simulate various processes. A preview of these models can be found in the WinSSS software that can be downloaded from Steve Krone's webpage.
  • video photo
    Nicolas Lanchier
    As a warming up, we will start with a brief overview of the main results about the voter model: clustering versus coexistence, cluster size and occupation time. The voter model is an example of interacting particle system - individual-based model - that models social influence, the tendency of individuals to become more similar when they interact. Each vertex of the lattice is characterized by one of two possible competing opinions and updates its state at rate one by mimicking one of its neighbors chosen uniformly at random. We will conclude with recent results about the one-dimensional Axelrod model which, like the voter model includes social influence, but unlike the voter model also accounts for homophily, the tendency of individuals to interact more frequently with individuals who are more similar. In the Axelrod model, each vertex of the lattice is now characterized by a culture, a vector of F cultural features that can each assumes q different states. Pairs of neighbors interact at a rate proportional to the number of cultural features they have in common, which results in the interacting pair having one more cultural feature in common.
  • video photo
    Steve Krone
    We will work through some of the basic ideas involved in modeling various types of interactions in spatial population biology using interacting particle systems (sometimes referred to as stochastic cellular automata). Some of the essential ingredients and behaviors come from simple models like the contact process and the voter model. These components can be combined and tweaked to obtain models with more biological detail, including epidemic behavior for host-pathogen systems, the spread of antibiotic resistance genes, etc. These models can be informative since real biological populations exhibit a high degree of spatial structure and this structure affects the interactions between individuals and species in ways that can dramatically alter dynamics compared to well-mixed systems. The computer exercises will allow students to alter some existing MATLAB code to simulate various processes. A preview of these models can be found in the WinSSS software that can be downloaded from Steve Krone's webpage.
  • video photo
    Nicolas Lanchier
    As a warming up, we will start with a brief overview of the main results about the voter model: clustering versus coexistence, cluster size and occupation time. The voter model is an example of interacting particle system - individual-based model - that models social influence, the tendency of individuals to become more similar when they interact. Each vertex of the lattice is characterized by one of two possible competing opinions and updates its state at rate one by mimicking one of its neighbors chosen uniformly at random. We will conclude with recent results about the one-dimensional Axelrod model which, like the voter model includes social influence, but unlike the voter model also accounts for homophily, the tendency of individuals to interact more frequently with individuals who are more similar. In the Axelrod model, each vertex of the lattice is now characterized by a culture, a vector of F cultural features that can each assumes q different states. Pairs of neighbors interact at a rate proportional to the number of cultural features they have in common, which results in the interacting pair having one more cultural feature in common.
  • video photo
    Sebastian Schreiber
    All populations experience stochastic
    uctuations in abiotic factors such as temperature, nutrient avail-
    ability, precipitation. This environmental stochasticity in conjunction with biotic interactions can facilitate
    or disrupt persistence. One approach to examining the interplay between these deterministic and stochastic
    forces is the construction and analysis of stochastic di erence equations and stochastic di erential equations.
    Many theoretical biologists are interested in whether the models are stochastically bounded and persis-
    tent. Stochastic boundedness asserts that asymptotically the population process tends to remain in compact
    sets. In contrast, stochastic persistence requires that the population process tends to be epelled" by some
    "extinction set". Here, I will review recent results on both of these proprieties are reviewed for models
    of multi-species interactions and spatially-structured populations. Basic results about random products of
    matrices, Lyapunov exponents, stationary distributions, and small-noise approximations will be discussed.
    Applications include bet-hedging, coexistence via the storage e ect, and evolutionary games in stochastic
    environments.
  • video photo
    Louis Gross
    This set of lectures and discussions will provide a quick one-day conceptual overview of stochastic issues in biology. Time permitting, I will point out the major conceptual approaches to stochasticity as typically applied in biology (random walks, Markov chains, birth and death processes, branching processes, agent-based models, stochastic DEs, diffusion processes, statistical modeling, Bayesian methods) and make the connection between these and deterministic analogs.

    The learning objectives for this day are:

    -Assist attendees in developing some intuition concerning how to think about biology from the perspective of probability distributions;
    -Encourage attendees to realize that there are diverse methods and models that can be applied across many fields of biology that have similar mathematical underpinnings, and these may be related to simpler deterministic models; and
    -Provide attendees with some hands-on experience with analysis of a stochastic process using simple computer tools.
  • video photo
    Linda Allen
    A brief introduction is presented to modeling in stochastic epidemiology. Several
    useful epidemiological concepts such as the basic reproduction number and the nal size
    of an epidemic are de ned. Three well-known stochastic modeling formulations are in-
    troduced: discrete-time Markov chains, continuous-time Markov chains, and stochastic
    di erential equations. Methods for derivation, analysis and numerical simulation of the
    three types of stochastic epidemic models are presented. Emphasis is placed on some of
    the di erences between the three stochastic modeling formulations as illustrated in the
    classic SIS (susceptible-infected-susceptible) and SIR (susceptible-infected-recovered)
    epidemic models. In addition, some of the unique properties of stochastic epidemic
    models, such as the probability of an outbreak, nal size distribution, critical commu-
    nity size, and expected duration of an epidemic are demonstrated in various models of
    diseases impacting humans and wildlife.
  • video photo
    Edward Allen
    Properties of the Wiener process are reviewed and stochastic integration is explained. Stochastic differential equations are introduced and some of their properties are described. Equivalence of SDE systems is explained. Commonly used numerical procedures are discussed for computationally solving systems of stochastic differential equations. A procedure is described for deriving Itˆo stochastic differential equations from associated discrete stochastic models for randomly-varying problems in biology. The SDEs are derived from basic principles, i.e., from the changes in the system which
    occur in a small time interval. Several examples illustrate the procedure. In particular, stochastic differential equations are derived for predator-prey, competition, and epidemic problems.

View Videos By