FIGURE 1.

Overview of the THD method's inputs, estimation, and modeling steps. THD estimation is a special case of kernel density estimation where the observations are haplotypes and the kernel bandwidth b is calibrated relative to a timescale parameter t ₅₀ defined as the median of the kernel distribution. The explicit estimation formula emphasizes that THD estimation has O (n ²) space and time complexity. THD values, used as proxies for the epidemic success of each haplotype, are typically regressed on a set of predictor variables. TMRCA, time to the most recent common ancestor between two haplotypes. ϵ is the error term of the regression model