Processes 2018, 6,56
The workflow characterization in Figure 4a is similar to that for IV, except that the Model
Mechanism descriptions (light blue box) are distinct in three ways. (1) Descriptions of entities and
activities are discretized sufficiently to specify in software a virtual analog of the Model Mechanism
that is faithful to details in III (e.g., see [33,34]). (2) Evidence is presented that the entities and activities
of the virtual analog are biomimetic. (3) The working hypothesis is that organized operation of
software entities and activities will be capable of generating a biomimetic phenomenon.
To achieve computational efficiencies and/or fine grain details, such as receptor trafficking and
molecular diffusion, influences of some entities and activities within the larger Model Mechanism are
often described using a combination of rules and continuous mathematics, as in V, rather than being
implemented as discrete biomimetic entities and activities. Doing so causes the software mechanisms
during execution to fall short of the definition of mechanism [35]. Nevertheless, an accurate descriptor
of the work product is a Simulation of a Model Mechanism. The following are examples.
Example VI.1: Simulations of Model Mechanisms are being used to help design and improve
therapeutic interventions in disease [36–38]. For example, they are providing improved
insight into possible failure modes of current treatments strategies for Tuberculosis (TB).
Building on their multilevel, multi-attribute Model Mechanism of an immune response
to TB, Linderman et al. [34] explored simulations of consequences of potential new
pharmacological interventions on six different model entities and activities, including
simulating immunomodulation by a cytokine; the consequences of oral and inhaled
antibiotics; and the effect of vaccination. In line with the features of a biological
mechanism (Table 1), their Model Mechanism identifies a phenomenon, the immune
response of TB as indicated by granuloma formation and function. Components are
represented at different spatial and temporal scales describe, starting with an agent-based
analogy of cell behavior (macrophages and T cells) across a cross-section of lung tissue.
Through rule-based probabilistic interactions, cell behavior is simulated in response
to a bacterial environment. At the lowest levels of simulation hierarchy, ordinary
differential equations were solved within each cell agent to simulate receptor/ligand binding,
trafficking, and intracellular signaling. Partial differential equations were solved to simulate
consequences of molecular diffusion. By linking their Simulation of a Model Mechanism for
TB to ordinary differential equation-based pharmacokinetic and pharmacodynamic models,
the authors simulated plausible consequences of the Model Mechanism’s behavior during
exposure to antibiotics. While simulations rely on some model compartments that are
Analogous-mechanism Models, the whole system is arguably a Model Mechanism. It is
biomimetic and represents an interconnected biological mechanism of granuloma formation
and immune response that extends from molecular to organ levels.
Example VI.2: A decade ago, several laboratories sought improved models of explanation
for vascular patterning defects observed in diabetic retinopathy and tumor angiogenesis.
Evidence suggested that an explanatory mechanism would involve disruption of
(1) Notch-driven specialization of endothelial cells into leading tip cells and following
stalk cells and (2) a feedback loop that links VEGF-A tip cell induction with delta-like
4 (Dll4)-notch-mediated lateral inhibition. Bentley et al. [39] constructed a hierarchical
Simulation of a Model Mechanism to explore the phenomenon of angiogenesis by
connecting Analogous-mechanism Models of these processes into a large biomimetic system.
The components included endothelial cell agents and membrane agents with multiple
cell agents arranged as a cylindrical capillary with each cell having membrane agents
distributed at the periphery. The study explored how different simulated VEGF environments
and filopodia dynamics would affect simulations of Notch-mediated selection of tip cells.
A staged simulation (temporally and spatially) first relied on a rule-based evaluation of
membrane processes for filopodium retraction or extension or notch response to VEGF.
195
The workflow characterization in Figure 4a is similar to that for IV, except that the Model
Mechanism descriptions (light blue box) are distinct in three ways. (1) Descriptions of entities and
activities are discretized sufficiently to specify in software a virtual analog of the Model Mechanism
that is faithful to details in III (e.g., see [33,34]). (2) Evidence is presented that the entities and activities
of the virtual analog are biomimetic. (3) The working hypothesis is that organized operation of
software entities and activities will be capable of generating a biomimetic phenomenon.
To achieve computational efficiencies and/or fine grain details, such as receptor trafficking and
molecular diffusion, influences of some entities and activities within the larger Model Mechanism are
often described using a combination of rules and continuous mathematics, as in V, rather than being
implemented as discrete biomimetic entities and activities. Doing so causes the software mechanisms
during execution to fall short of the definition of mechanism [35]. Nevertheless, an accurate descriptor
of the work product is a Simulation of a Model Mechanism. The following are examples.
Example VI.1: Simulations of Model Mechanisms are being used to help design and improve
therapeutic interventions in disease [36–38]. For example, they are providing improved
insight into possible failure modes of current treatments strategies for Tuberculosis (TB).
Building on their multilevel, multi-attribute Model Mechanism of an immune response
to TB, Linderman et al. [34] explored simulations of consequences of potential new
pharmacological interventions on six different model entities and activities, including
simulating immunomodulation by a cytokine; the consequences of oral and inhaled
antibiotics; and the effect of vaccination. In line with the features of a biological
mechanism (Table 1), their Model Mechanism identifies a phenomenon, the immune
response of TB as indicated by granuloma formation and function. Components are
represented at different spatial and temporal scales describe, starting with an agent-based
analogy of cell behavior (macrophages and T cells) across a cross-section of lung tissue.
Through rule-based probabilistic interactions, cell behavior is simulated in response
to a bacterial environment. At the lowest levels of simulation hierarchy, ordinary
differential equations were solved within each cell agent to simulate receptor/ligand binding,
trafficking, and intracellular signaling. Partial differential equations were solved to simulate
consequences of molecular diffusion. By linking their Simulation of a Model Mechanism for
TB to ordinary differential equation-based pharmacokinetic and pharmacodynamic models,
the authors simulated plausible consequences of the Model Mechanism’s behavior during
exposure to antibiotics. While simulations rely on some model compartments that are
Analogous-mechanism Models, the whole system is arguably a Model Mechanism. It is
biomimetic and represents an interconnected biological mechanism of granuloma formation
and immune response that extends from molecular to organ levels.
Example VI.2: A decade ago, several laboratories sought improved models of explanation
for vascular patterning defects observed in diabetic retinopathy and tumor angiogenesis.
Evidence suggested that an explanatory mechanism would involve disruption of
(1) Notch-driven specialization of endothelial cells into leading tip cells and following
stalk cells and (2) a feedback loop that links VEGF-A tip cell induction with delta-like
4 (Dll4)-notch-mediated lateral inhibition. Bentley et al. [39] constructed a hierarchical
Simulation of a Model Mechanism to explore the phenomenon of angiogenesis by
connecting Analogous-mechanism Models of these processes into a large biomimetic system.
The components included endothelial cell agents and membrane agents with multiple
cell agents arranged as a cylindrical capillary with each cell having membrane agents
distributed at the periphery. The study explored how different simulated VEGF environments
and filopodia dynamics would affect simulations of Notch-mediated selection of tip cells.
A staged simulation (temporally and spatially) first relied on a rule-based evaluation of
membrane processes for filopodium retraction or extension or notch response to VEGF.
195