Our Model: AQUAINT


AQUIRE

Introduction

A foundational understanding of how synthetic biology can be integrated into fluid dynamic environments is critical for advancing both theoretical and applied aspects of the field. Biological systems are inherently complex, and when introduced into environments governed by fluid flow—such as pipes, open water, or engineered surfaces—their behavior becomes even more difficult to predict. To address this challenge, a robust and modular mathematical framework is required to abstract and simulate the interactions between synthetic constructs and their surrounding fluid environments. Such a framework enables researchers to assess whether a construct can function reliably and sustainably under specific physical conditions.

The challenge lies in the vast diversity of synthetic biology applications and the multitude of possible deployment scenarios. Each use case—whether targeting microbial communities, therapeutic delivery, or environmental remediation—presents unique constraints. Additionally, incorporating fluid dynamics into these models is essential for realism, but doing so while preserving the ecological and community-level dynamics affected by genetic engineering introduces significant complexity. The interplay between engineered function and environmental transport must be carefully balanced to avoid oversimplification or loss of biological fidelity.

To begin addressing these challenges, we developed AQUAINT, a modular framework of novel mathematical models tailored to distinct fluid dynamic contexts. These include scenarios such as phage transport through plumbing systems, bacterial dispersal in flowing aquatic systems while preserving community interactions, and the deployment of biofilms onto various steel surfaces. Each model captures key aspects of fluid transport, biological persistence, and interaction dynamics, offering a foundation for predictive design and deployment. These models allow us to identify the key design principles for genetic engineering and effective treatment, guide the development of constructs, and evaluate how well a given chassis can function within its specific aquatic environment. With this framework, we developed novel mathematical models to describe the deployment of our chassis in aquatic environments, further extracting information of how the chassis may behave.

Additionally we incorporate the usage of our machine learning classifier AQUIRE to determine how biotic, abiotic, and mechanical factors of systems affect our chassis’ functionality in a given system.

Ultimately, integrating fluid dynamic principles into synthetic biology modeling of aquatic systems is not only essential for scientific rigor but also for improving the efficiency, reliability, and scalability of applications concerned with synthetic biology.

Open Moving Water


Chassis Transport

For transport of a synthetic biology chassis in running water, the transport can be described by a partial differential equation governing its time evolution as it is transported through advection and diffusion within a fluid environment. Given the computational complexity of computational fluid dynamics simulations using Navier-Stokes equations, a good initial test of feasibility begins with partial differential equations representing advection, diffusion, and reaction through the transport equation incorporated into the various dynamics.

Equation

In the above PDE, C represents the concentration of the chassis, and u is the velocity of the fluid. The necessity of a diffusion term can be calculated through the Peclet number, given by:

Equation

where u is the averaged flow velocity, L is the characteristic length in the direction of the flow velocity, and D is the molecular diffusion coefficient of the chassis. When the Peclet number is extremely high, it indicates that the model can be approximated by ignoring diffusion, easing simulations.

The various other elements important to the system can be modelled similarly. For example, if S is a concentration dynamically related to C:

Equation

To initialize the system, we represented both the chassis and the target population using Gaussian spatial distributions. Each population was centered at distinct positions along the domain to reflect an initial separation between the engineered construct deployment and its intended target. The variance (σ) of each Gaussian was chosen to capture differences in spatial extent: the target population was assigned a broader distribution to represent its dominance in the environment, while the chassis population was initialized with a narrower distribution to reflect localized deployment. Here xc denotes the deployment location of the chassis in space and xt denotes the centered point in space of the target concentration which the chassis will affect:

Equation

This transport framework applies broadly to any synthetic chassis deployment where it is being transported in a moving water system.

By explicitly incorporating advection and diffusion into chassis design, synthetic biologists can better predict how constructs disperse, persist, and interact with surfaces or communities. Furthermore, they can identify engineering targets that improve performance in specific environments, such as adhereability of the construct. Through this framework, synthetic biologists can predictably optimize deployment strategies by matching chassis properties to flow regimes, geometries, and system contexts. Additionally, they can forecast chassis performance to determine whether their chassis can carry out its desired function. The advection-reaction-diffusion approach where u represents a streamline of the flow is a good starting point in the construction of models that incorporate fluid dynamics through computational simulation of Navier-Stokes equations for viscous flow.

Pipes & Plumbing


Chassis Transport Equation

The transport of a synthetic biology chassis through a plumbing system, where it reacts with a surrounding substance, can be described by a partial differential equation governing its temporal evolution as it disperses or diffuses within the fluid. For example, the chassis concentration can be modeled as a function of the radial coordinate under the assumptions of azimuthal and translational symmetry to understand its ability to be correctly transported within a system. This framework captures both passive transport (diffusion, advection) and active interactions (adsorption, reaction, or colonization).

Equation

In many engineered environments, eddy diffusion dominates molecular diffusion. We approximate this using standard chemical engineering relations, where the dimensionless constant α characterizes the flow regime, u is the averaged bulk flow speed, and R is the pipe radius (e.g., α = 0.05 in pipe flow; Venkatram & Weil, 2021).

Equation

At the bulk–surface interface (e.g., biofilm, tissue, or material surface), reaction rates govern how the chassis interacts with its environment. This can be incorporated through the following mixed boundary condition,

Equation

The relevant timespan is determined by the interaction window between the flowing chassis and the target surface. This can be calculated from the fluid velocity u and the characteristic length of the surface region, L, given by the length of the reactive surface.

Equation

The system can be initialized at various chassis concentrations, assumed spatially uniform, and scaled to the geometry of the system.

Equation

This transport framework applies broadly to any synthetic chassis deployment into a plumbing environment or pipe system (household plumbing, tubes in medical constructs, etc.). By explicitly incorporating fluid dynamics into chassis design, synthetic biologists can better predict how constructs disperse, persist, and interact with surfaces or communities. Furthermore, they can identify engineering targets that improve performance in specific environments, such as adhereability of the construct. Through this framework, synthetic biologists can predictably optimize deployment strategies by matching chassis properties to flow regimes, geometries, and ecological contexts.

Material Attachment


For deployment of a synthetic biological material, we reduce the complexities of the system so that the material is able to be simply modeled by shear and normal detachment through an assumption of homogeneity.

Due to the fluid dynamics that interact with the material, we use simple equations of stress and strain and contact angles, θ, to understand the variable deployment positions that favor sustainability. This composes our shear, τ, and normal, σn stress equations into the following dependent on the aquatic environment, where L is the material thickness, ρ is the fluid density, μ is the viscoelasticity, and u is the speed of the fluid:

Equation

We can then look to see the various contact angles and properties of water where deployment is possible or impossible. Additionally, we can determine the critical shear stresses and critical normal stress of a given material to find out what genetic design principles need to be applied to improve various elements of the construct. Here, ks and kn are detachment coefficients and n and m characterize the level of detachment after reaching the threshold detachment value:

Equation

Additionally, when a construct is deployed into an environment, various other factors can affect the surface structure, such as ion degrading a surface or other solutes reacting with the surface. This can be incorporated into the equations using a Michaelis-Menten type of reaction following this scheme, where S is the reactive concentration in the aquatic environment and E is the interface of the material:

Equation

With these kinetics in mind, a preliminary model of deployment can be made to discover the various aquatic conditions and environments where deployment is feasible or infeasible. With this framework of model, genetic design principles can be made to understand the environments where a reactive substance or the fluid properties cause the construct to not sustain its deployment.

Categories of Aquatic Systems


AQUIRE-ING MACHINE LEARNING INTO FEASIBILITY ANALYSIS

Equation

With a foundational understanding of how the various deployment techniques and situations can be treated, a commonly overlooked aspect of synthetic biology is the natural compatibility of a bacteria to sustain itself in the environment into which it is deployed.

Therefore, a predictive model is needed to characterize a construct’s capacity for self-sustainability, enabling its design to be tailored to specific environments and ensuring effective deployment where it can persist. For our model to be all encompassing of the ability and predictive aspect of synthetic biology constructs in model, we couple our database model AQUIRE to incorporate the biotic and abiotic factors that an ecosystem would imply on a chassis’ success.

We do this through an adjustment of the constructs carrying capacity (K) and growth rate (α) based on the fitness parameter assigned by AQUIRE based on the following mapping where γ is the fitness parameter of the chassis:

Equation

To demonstrate the importance of this parameter for predictive approaches of chassis deployment, we first look at the differences of behavior of Acinetobacter under differing γ values.

Equation

At gamma values below 0.5, the HAB remains above the threshold required for full suppression. Although the chassis functionally reduces HAB levels, key deterrents such as microcystins are unsolved and persist into further waterways. This indicates that synthetic biologists if willing to deploy in a less fit environment must compromise or engineer further their construct to be specifically effective for a given environment.

We additionally applied this transformation through a gamma term onto the B. Subtilis deployment model to better understand the environments where it can sustain itself properly.

Equation

We additionally see that the ability of the B. Subtilis biofilm depends greatly on its level of fitness in an environment. If its fitness isn't high enough, it will not sustain itself and will thus be dysfunctional within its environment.

While the precise link between chassis fitness and environmental performance remains unclear, we incorporate this scalar into our model to have a predictive understanding of the deployability of a chassis.

The feasibility of any synthetic biology chassis depends not only on its engineered traits but also on its ecological fitness within the deployment environment. AQUIRE’s machine‑learning–derived fitness parameter allowed us to quantify this compatibility and incorporate it into predictive simulations, making the modeling both more realistic and more actionable. This combined framework ensures that future chassis design can be tailored not just for function, but for persistence and effectiveness in their targeted real‑world systems.

References