Extended Data Fig. 5: Sampling estimates of transmission events and sources of infections.

(a-c) The sampling cascade of transmission events was modelled by comparing the number of phylogenetically reconstructed source–recipient events to the estimated number of infection events under the incidence model, by gender, age band and survey round of infected individuals. Throughout, shown are the number of sampled and unsampled transmission events, the estimated proportion of transmission events that were ever deep-sequenced, and log ratios of estimated proportion of transmission events that were ever deep-sequenced in any strata relative to the overall average across strata (point estimates: dots, 95% confidence intervals: linebars). Estimates are based on n = 227 source–recipient pairs and n = 1,117 observed incidence events. (d-f) Additional differences in source sampling were modelled by considering unsuppressed individuals as potential sources of infection, and calculating the number of unsuppressed individuals in a round that were ever deep-sequenced. Throughout, shown are the number of sampled and unsampled possible transmission sources, the estimated proportion of possible sources that were ever deep-sequenced, and log ratios of estimated proportion of possible sources that were ever deep-sequenced in any strata relative to the overall average across strata (point estimates: dots, 95% confidence intervals: linebars). Estimates are based on n = 227 source–recipient pairs and n = 3,265 participants with HIV and measured viral load.