FIG 3.

Additional results from application of the beta-binomial sampling method to the simulated data set. (A) Probability of a donor-identified variant being either transferred or observed as transferred (“called”) in a recipient host, as a function of donor variant frequencies. The observed probabilities of donor-identified variants being called in a recipient host are shown in black, calculated directly from the simulated data set using 3% frequency bins. The 95% confidence intervals assume that the probability of variant transfer follows a binomial distribution with the number of trials being the number of donor-identified variants present in a frequency bin and the success probability given by the calculated probability of transferred variants observed in the frequency bin. Probabilities of donor-identified variants being truly present in a recipient host are shown in purple, given bottleneck size estimates from the beta-binomial sampling method. Probabilities of donor-identified variants being called present in a recipient host are shown in gray, given bottleneck size estimates from the beta-binomial sampling method. (B) N_b estimates for simulated data sets that differ in coverage levels. At each coverage level, 5 data sets were generated under the same parameters and assumptions as those for the data set shown in Fig. 2A. Both the exact beta-binomial sampling method and the approximate version of this method were used to estimate N_b for each data set. N_b maximum likelihood estimates and 95% confidence intervals are shown in purple for the exact beta-binomial sampling method and in pink for the approximate method.