Interaction-driven Spatial Agent-based Models at Multiple Levels of Biological Organization.
Ferreira, Giordano.
2019
-
Most biological
systems are hard to investigate in the laboratory or in the field. Computational
modeling has been used to help researchers to generate novel hypotheses to be tested
empirically. Computational simulations allow researchers to perform a large number of
experiments with relatively low effort. Agent-based modeling (ABM) has emerged as a
viable modeling paradigm in biology. ABM is ... read morea bottom-up approach in which the phenomena
of interest emerge from behavioral rules of the minimal component of the model: the
agent. Some agent-based models can be described as spatial. In these models, the area in
which agents behave (e.g., move) is described explicitly. Agent-based models can also be
described as interaction-driven. In these models agents are capable of performing a
simulated interaction with each other. Interaction is a type of behavior in which agents
exchange some kind of information. ABMs for biology exist at different scales, from the
molecular scale in which agents are molecules inside a cell to the macroscopic scale in
which agents are animals in some social context. This dissertation focuses on
Interaction-driven Spatial Agent-based Models for biology at different scales. It is
divided into three parts. Part I describes the Interaction-driven Spatial Agent-based
Model Framework (IS-ABM Framework), a standard way to plan agent-based models for
biological systems. The IS-ABM Framework divides the models into four components: agent,
environment, interaction and scheduler. For each component, a list of questions that
must be answered by the modeler is presented. The goal of the IS-ABM Framework is to
help newcomers to the ABM community with the modeling procedure. This dissertation also
presents a review of eleven papers proposed for gathering an initial pool of examples
for newcomers to the IS-ABM Framework. This dissertation also presents two
Interaction-driven Spatial Agent-based Models for two distinct biological systems. Part
II focuses on understanding social behaviors of animals. Specifically, it investigates
territory exploration tasks in which individuals must visit checkpoints distributed
across an environment. Checkpoints have a quality that can be sensed by individuals.
Individuals must balance the quality of the checkpoints with the costs to reach them.
The main instance explored here is the mating selection task. This dissertation presents
a biologically plausible model for the mating selection task in treefrogs. It also
investigates the extension to cases when individuals end up visiting undesirable
checkpoints by accident. Part III presents a proof-of-concept model to explore the
regeneration process in Planaria. Even though biologists know genes involved in some
parts of the regeneration process, it is necessary to have an algorithmic explanation of
the process. Specifically, how morphological information can be stored and edited
without genetic modifications after the full development of the organism. This
dissertation presents a cell-to-cell communication mechanism that dynamically discover
an organisms shape and uses this distributed information for regeneration. This
mechanism replicates various regeneration capabilities displayed by Planaria such as
random cell death, tissue removal, reliability against noise, stem cell migration and
proliferation.
Thesis (Ph.D.)--Tufts University, 2019.
Submitted to the Dept. of Computer Science.
Advisor: Matthias Scheutz.
Committee: Anselm Blumer, Michael Levin, and Hiroki Sayama.
Keywords: Computer science, and Systematic biology.read less - ID:
- th83m994b
- To Cite:
- TARC Citation Guide EndNote
