Figure 4. Bayesian posteriors for the age effect from negative binomial models.

These plots show the priors and posteriors for from a negative binomial model fitted to the number of crossovers called by duoHMM in either informative duos only (a–d) or both informative and partially informative duos (e–h), with either the same (Models 2 and 2*) or different (Models 2.2 and 2.2*) age effects for each cohort (see Methods). We have plotted since, under the negative binomial model, the expected number of crossovers increases by this factor per year. The axes have been chosen to facilitate comparison with the normal model shown in Figure 3, assuming a baseline of 38, that is, the bounds [0.985,1.013] on under the negative binomial model are approximately equivalent to the bounds [38 × log(0.985), 38 × log(1.013)]=[−0.57,0.49] on β_age under the normal model.