Extended Data Fig. 3: Individual genotype plots.

Six heat maps showing the evolution of distinct genotypes for individual model runs. Each heat map shows genotype frequency, from 2018 to 2039, for all genotypes that rose above a 0.001 frequency at any point during this period. Frequencies below 0.001 are shown in white, whereas genotype frequencies above 0.001 are shown according to the colorbar on the right. The seven-letter genotype codes refer to, from left to right: the K76T locus, the N86Y locus, the Y184F locus, a second copy of N86Y, a second copy of Y184F, the R561H locus, and the presence (‘2’) or absence (‘1’) of a combination of piperaquine-resistant alleles. The top two heat maps show two different model runs for the status quo – differing only in that the model’s behavior is stochastic – where 3-day artemether-lumefantrine is maintained as first-line therapy. The outcomes of these runs differ, with different double-resistant (to AL) genotypes emerging and spreading. The four major AL-resistant genotypes that spread in these model runs are labeled with their efficacies in the heatmap (under a 3-day course of AL). These four double-resistant genotypes are labeled with asterisks in the four heatmaps at bottom, showing that some of the management strategies evaluated in this modeling exercise are successful at delaying the onset of AL double-resistance. For the four strategies shown at bottom, a typical model run was chosen to display. Note that under MFT, more genotypes emerge but selection pressure is weaker on each individual genotype.