Figure 4. Model-predicted efficacy of precision-guided sterile insect technique (pgSIT) egg releases on Ae. aegypti population suppression as a function of release scheme, male sterility, and female viability.

Weekly releases were simulated in a randomly mixing population of 10,000 adult mosquitoes using the MGDrivE simulation framework² and parameters described in Supplementary file 1l. Population suppression outcomes were identified as being most sensitive to model parameters describing the release scheme, male fertility, and female viability (Figure 4—figure supplement 1). (A) These parameters were varied in factorial experiments assessing suppression outcomes including probability of elimination (the percentage of 60 stochastic simulations that result in Ae. aegypti elimination), and window of protection (the time duration for which ≥50% of 60 stochastic simulations result in ≥90% suppression of the Ae. aegypti population). Female viability was varied between 0 (complete inviability) and 0.15, male fertility was varied between 0 (complete sterility) and 0.15, release size was varied between 0 and 500 eggs released per wild adult, and the number of weekly releases was varied between 0 and 52. Regions of parameter space for which the probability of elimination exceeds 90% are depicted in purple, and in which the window of protection exceeds two years in light blue. Time-series of population dynamics for select parameter sets are depicted in a–h. Here, the total female population is denoted in red, and the Cas9-carrying female population is denoted in blue. The light blue shaded region represents the window of protection. Imperfect female inviability and male sterility result in lower probabilities of elimination; however the window of protection lasts for several years for male fertility and female viability in the range 0–0.15 for simulated release schemes. (B) Regions of parameter space for which the probability of elimination exceeds 90% are depicted as a function of male fertility (x-axis), female viability (y-axis), and the minimum number of weekly releases required to achieve this (shadings, see key). Release size is set to 250 eggs per wild adult. The shaded square depicts the region of parameter space in which male fertility is between 0% and 10% and female viability is between 0% and 10%. A ≥90% elimination probability is achieved with ~20–32 weekly releases for pgSIT systems having these parameters. Source data are provided in Supplementary file 1.

Figure 4—figure supplement 1. Sensitivity of precision-guided sterile insect technique (pgSIT) population suppression outcomes to model parameters.

(A) Partial dependency (thin traces) and individual conditional expectation plots³ (magenta dashed lines) are depicted for regression models fitted to three model outcomes: (1) probability of elimination (POE), (2) window of protection (WOP) (measured in years), and (3) reduction in cumulative potential for transmission (RPT). These three outcomes were evaluated for several parameters – number of releases, size of releases (as a proportion of the number of wild adults in the environment), release interval (in days), gRNA cutting rate, Cas9 maternal deposition rate, male fertility, male mating competitiveness (relative to wild-type males), and female viability. Population suppression outcomes are most sensitive to release schedule parameters (number, size, and interval of releases), male fertility, and female viability. (B) Statistical sensitivity analysis metrics on the simulated data are compared to importance metrics for surrogate models. For the first-order statistical sensitivity analysis, we used Delta⁴, Fourier Amplitude Sensitivity Test (FAST)⁵, and High-Dimensional Model Representation (HDMR)⁶; while for the emulators we used Permutation feature importance⁷, and SHapley Additive exPlanations (SHAP)⁸. The normalized importance barcharts are in good agreement with the variable-to-outcome rankings (Supplementary file 1l, m). Source data are provided in Supplementary file 1.

Figure 4—figure supplement 2. Sensitive and rapid detection of transgenic DNA fragments with the sensitive enzymatic nucleic acid sequence reporter (SENSR) assay.

(A) Intensity of signal from SENSR assay for both Cas9 and gRNA^dsx,ix,βTub constructs along a concentration gradient, as ratio of pgSIT^♀Cas9 to wild-type (WT) mosquitoes, and No Template Control (NTP). The fluorescence signal represents the background-subtracted signal (n = 4). (B) Speed of cleavage along a ratio of pgSIT^♀Cas9 to WT mosquitoes, as target abundance. The speed of collateral activity is represented using an half-maximum fluorescence (HMF) analysis (n = 4) (Supplementary file 1o).