Fig 1.
Representative genomes of cohorts or individual mosquitoes within EMOD.
The model supports the inheritance of traits from parents (a), random mutations of alleles (b), and definitions of phenotypic traits associated with a combination of genes or alleles, that are expressed only when those combinations are present (c). Inheritance of genes is modeled as a Mendelian process, and combinations of alleles can be mapped to combinations of traits via a many-to-many mapping.
Fig 2.
Inheritance rules for gene drives.
In this example, ‘a1’ is the driven allele and ‘a0’ is the wild-type allele. Red mosquitoes represent mosquitoes with the driven allele ‘a1’ either in the homozygous or heterozygous configuration. Black mosquitoes represent wild-type mosquitoes, which are homozygous in ‘a0’. The drive successfully cleaves the target site 95% of the time. In the event of drive failure, the wild-type allele remains. This is akin to modeling drive failure in terms of the driven allele failing to home in to the target site. This does not include target site resistance through natural mutation or non-homologous end joining (NHEJ). No other mutations are considered. Three mating scenarios are presented: (A) When a drive carrying heterozygous mosquito mates with a homozygous wild-type mosquito, (B) a heterozygous-with-drive mosquito mates with a homozygous-with-drive mosquito, and (C) two heterozygous-with-drive mosquitoes mating.
Fig 3.
Seasonality and spatial simulation setup using EMOD.
(A) Single peak seasonality profile characteristic of a Sahelian seasonal and transmission setting. All simulations presented here with a prevalence endpoint measure use this seasonality profile resulting in an annual EIR of around 125 infectious bites per person. (B) Spatial grid used for spatiotemporal gene drive simulations. Six nodes with the largest human population are selected as release sites for genetically modified mosquitoes carrying drives and are marked in red.
Fig 4.
Example of how traits and alleles interact in the vector genetics model.
There is no seasonal variation and no spatial component. (A) Genomes of mosquitoes in the model. Mosquitoes homozygous in ‘a0’ are the wild-type mosquitoes. Male mosquitoes carrying the ‘a1’ allele in a homozygous or heterozygous configuration have decreased mortality, which is thus a dominant trait, and is modeled as a halving of their probability of dying. (B) Distributions of genomes in the population over time average over 50 stochastic realizations. Male mosquitoes homozygous in the ‘a1’ allele are released mid-year during the first year of the simulation.
Fig 5.
Example of species introgression in the vector genetics model.
The mean (solid line) and one standard deviation (shaded area) of genomes in the population over time of 50 stochastic realizations when 10000 male and female mosquitoes of species 2 that are homozygous in ‘b0’ are introduced into a population containing only species 1 mosquitoes homozygous in ‘a0’ 200 days after simulation starts. (A) When no introgression occurs, and each species has equal habitat to procreate. (B) Hybrid mosquitoes heterozygous in ‘a0’ and ‘b0’ have a 10% lower mortality rate than species 1 or species 2 mosquitoes.
Fig 6.
Example of how gene drives behave in the vector genetics model.
The mean (solid line) and one standard deviation (shaded area) of genomes in the population over time of 50 stochastic realizations when male mosquitoes with drive are released into a wild-type population. There is no seasonal variation and no spatial component. Mosquitoes homozygous in ‘a0’ are the wild-type mosquitoes. Male mosquitoes carrying the ‘a1’ allele in a homozygous configuration are released 200 days after simulation starts. The ‘a1’ allele is a driven allele and has a homing rate of 50% to the ‘a0’ site. (A) There are no fitness costs associated with drive mosquitoes. (B) Mosquitoes with drive in either a homozygous or heterozygous configuration have a 10% higher mortality rate than wild-type mosquitoes.
Fig 7.
Tracking the effect of insecticide resistance on prevalence in a Sahelian setting.
RDT prevalence (top row) across a period of six years averaged over 50 stochastic realizations for three scenarios—when no interventions are deployed, when ITNs are deployed at 60% coverage absent any resistance, and with the presence of resistance. Establishment rates of susceptible and resistant genomes in the scenario with resistance (bottom row). The shaded area around the mean represents one standard deviation calculated across the 50 stochastic realizations. ITNs are distributed every three years at the beginning of the peak season.
Fig 8.
Likelihood of elimination in a Sahelian setting using gene drives that reduce the probability of transmission of parasites from mosquitoes to humans.
Three scenarios are presented—one without ITNs, and ITNs distributed every three years over a six year period at 60% and 80% coverage, respectively. The fraction of simulations eliminating is evaluated over 50 stochastic realizations for a given value pair of probability of transmission from mosquito to human and likelihood of successful gene drive of the gene responsible for reduced transmission.
Fig 9.
Evolution of true prevalence over time when different intervention packages are deployed in a multi location Sahelian setting.
Results are from a spatial simulation with 150 one square kilometer nodes with varying population and larval habitat sizes spread across 300 square kilometers. Four different scenarios are explored—absent any vector control interventions, release of genetically modified mosquitoes carrying a drive that reduces the probability of parasite transmission from mosquitoes to humans, deploying ITNs every three years just as the peak season begins to pick up, and a combination of gene drive and ITNs. The average of 50 stochastic realizations of each scenario is represented by solid lines while the shaded area represents one standard deviation.