Edit on GitHub

    Design

    GeneDrive.jl is predicated on composability. This system design principle allows specific user requirements to be met by mixing and matching modular components (or adding new ones). Because environmental variation - not to mention biological and genetic diversity - can be expressed in many forms, the "composability" of GeneDrive.jl means that the features described below may be selected, assembled, and augmented to develop many unique scientific explorations.

    Feature modularity in GeneDrive.jl extends to solution methods: the dynamic model is built on top of DifferentialEquations.jl, enabling users to experiment with the robust suite of solvers in that package. The decision model employs JuMP.jl and can be run using a plethora of free as well as paid solution algorithms for nonlinear and mixed integer problems. Once built, the same data model can be evaluated using either ODE or optimization approaches with no further customization required.

    Climate

    To enable experimentation with alternative environmental assumptions, three categories or "subtypes" of Temperature (°C) inputs are defined in the GeneDrive.jl data model:

    • ConstantTemperature creates a constant environment for the duration of a simulation: ConstantTemperature(27.0).
    • SinusoidalTemperature furnishes an idealized, seasonally fluctuating temperature regime according to user-supplied values for amplitude, periodicity, time period, and mean: SinusoidalTemperature(4.75, 2, 365, 20.75).
    • TimeSeriesTemperature is applicable for empirical data: TimeSeriesTemperature([vector_of_example_data]).

    Beyond increases in the average trend that can be accommodated within the three Temperature subtypes above, GeneDrive.jl accommodates heatwaves and cold snaps via TemperatureShockData that imposes time-bound increases or decreases in °C, e.g., augmenting the ambient temperature by 2°C between days 21 and 25 and again between days 37 and 42 in the location of interest: TemperatureShockData(node,[(21.0, 25.0),(37.0,42.0)],2.0).

    Biology

    Lifecycle dynamics vary among species thanks to population-specific vital rates and regulatory mechanisms as well as ecological and other factors. The Organism type in GeneDrive.jl is defined by the LifeStages and Genetics particular to a given study species, and is categorized by Species. To view the species already defined in the package, run:

    julia> ? Species

    GeneDrive.jl draws on the thermal biology literature to provide empirically-derived functions that characterize species-appropriate temperature responses. The following - named for the first author of the scientific publication from which they were sourced - are fully parameterized GeneDrive.jl data models:

    • stages_rossi() and stages_moustaid(), applicable to AedesAegypti
    • stages_abiodun(), applicable to AnophelesGambiae

    To permit experimentation with alternative field or laboratory-sourced vital rates, the pre-defined GeneDrive.jl data model stages_noresponse() - named to indicate that no functional response to temperature is assumed - can be populated with stage-specific information using package accessor functions. See API Reference.

    Stage-specific regulatory mechanisms and their contributions to population dynamics can also be explored. See Density models under the type DensityDependence.

    Genetics

    The Genetics type in GeneDrive.jl defines the likelihood with which an organism will produce offspring and pass on its genetic material. This likelihood is currently assumed to be equivalent for all wildtype organisms, following Mendelian inheritance principles. However, genetic-based methods of biocontrol - a category that ranges from actual modification to bacterial infection - alter natural inheritance patterns, with affected genotypes having a greater or lesser ability to propagate.

    Beyond this, Genetics data includes information unique to modified genotypes including fertility rates, sex ratios, and varied fitness costs (e.g. how mortality rates for those organisms are biased with respect to their wild counterparts). Together with the vital rates defined in LifeStages, Genetics data is used to create the Organisms that populate a GeneDrive.jl experiment. The following are fully parameterized data models included in the package:

    • genetics_mendelian(): For wildtypes.
    • genetics_ridl(): For the biocontrol technology RIDL (suppression), developed by Oxitec.
    • genetics_wolbachia(): For two biocontrol approaches (i.e., suppression and replacement) that infect organisms with Wolbachia (multiple existing strains).
    • genetics_mcr(): For a gene drive biocontrol technology (replacement) tested exclusively in laboratory environments.

    Geography

    Connectivity between geographic locations informs population dynamics by allowing organisms to migrate and spatially propagate genetic material. Study species defined by the GeneDrive.jl data model may inhabit a Node - a single homogenous habitat - or a heterogeneous Network. The Network is a collection of interconnected Nodes, each of which are geolocated by latitude/longitude coordinates and may exhibit environmental differences. See the data model tutorial for an example of how to construct both Node and Network geographies.

    Movement between all Node locations in a Network is uniquely specified for each species, life stage, and genotype using assign_migration!. See the data model tutorial for an example implementation. This feature enables users to account for a diversity of demographic and genetic migration tendencies as well as exogenous factors, facilitating scientific investigation in an important area.

    Interventions

    Biological control experiments are conducted in GeneDrive.jl by adding modified organisms at periodic intervals using the Release object. In the dynamic model, such interventions are defined by the user when solving the problem. In the decision model, feasible interventions are optimized according to the operational constraints and objective function defined by the user.

    When specifying an intervention schedule in the dynamic model, fixed or variably-sized releases can occur at flexible intervals by entering the applicable Node, Species,LifeStage, genotype index, and timing (day of desired release).

    • Fixed release size: Release(node, AedesAegypti, Female, 1, [46.0, 53.0, 60.0, 67.0], 100)
    • Variable release size:Release(node, AedesAegypti, Female, 1, [46.0, 53.0, 60.0, 67.0], [100, 150, 200, 250])

    Alternatively, interventions may be conducted in an adaptive manner that directly accounts for the real-time size of the standing wildtype population. This is done using the ProportionalRelease approach. As above, information defining the Node, Species,LifeStage, genotype index, and timing is entered. However, instead of an Int64 (count) specifying the release size, a Float64 denotes the relative magnitude of the intervention with respect to the population of interest, which may be adult Males, Females, or both ("All"). In the examples below, release_times = [46.0, 53.0, 60.0, 67.0]:

    • Proportional release, where release size is equivalent to standing population of Males at each release time: ProportionalRelease(node, AedesAegypti, Male, 1, release_times, 1.0, "Males")
    • Proportional release, where release size amounts to 50% of the combined standing population of Male and Female adults at each release time: ProportionalRelease(node, AedesAegypti, Male, 1, release_times, 0.5, "All")