ResistFlow: A Forward-Time Simulation Framework for Exploring Insecticide Resistance Evolution in African Malaria Vectors

Insecticide resistance in Anopheles malaria vectors threatens the effectiveness of vector control across sub-Saharan Africa. Genomic surveillance programmes such as MalariaGEN now provide allele frequency data across multiple African countries and time points, but these data are used predominantly to describe resistance that has already spread rather than to explore how it might spread further. We present ResistFlow, an open-source proof-of-concept framework that simulates resistance allele trajectories under configurable intervention scenarios. The framework implements a single-locus Wright-Fisher model with selection and genetic drift, initialises from real observed allele frequencies - from the MalariaGEN Ag3 API or from user-supplied data and estimates population-specific fitness parameters automatically from temporal observations. We validate against Vgsc L995F (kdr) trajectories in two independent An. gambiae populations. In Cameroon, where three time points including an intermediate observation are available, the framework recovers an identifiable point estimate and reproduces the observed trajectory shape. In Burkina Faso, where only a baseline and a fixation endpoint exist, the parameters are not identifiable; the framework detects this and reports a bounded estimate (s ≥ 0.150) rather than a misleading point value. The contrast between these cases yields a concrete recommendation for surveillance design: observations sampled mid-sweep constrain evolutionary parameters substantially more than repeated sampling after fixation. ResistFlow is available under the MIT License.