Effects of Darwinian Selection and Mutability on Rate of Broadly Neutralizing Antibody Evolution during HIV-1 Infection
Fig 3
The stage method is better than the time-bin method for estimating evolutionary rate dynamics.
(A) The evolutionary rate for the constant-rate (CR) dataset was estimated using both restricted and relaxed log-normal clock models (labeled CR_restricted_clock and CR_lognormal_clock respectively). The estimated mean evolutionary rates from the two models are in good agreement. The mean evolutionary rates of the decreasing rate (DR) dataset estimated from the MCC tree and the 1000 time scaled Bayesian trees (labeled DR_MCC_tree and DR_1000_tree respectively) are highly consistent with the expected rates (red dots) derived from the calculated evolutionary rate of the CR dataset (see Materials and Methods). (B) The mean evolutionary rates for the eight time bins of the DR dataset were estimated from a single MCC tree (blue) and the mean and the 95% highest probability density (HPD) intervals estimated from 1000 time scaled Bayesian trees (green). The estimated evolutionary rates for each bin are significantly different from the expected rate (red). (C) The expected rates (red) of the three stages of the DR dataset are within the estimated 95% HPD of the mean evolutionary rates (purple), suggesting the stage method is reliable for characterizing evolutionary rate changes over time.