Table 2.

Estimation of selection acting on gene methylation state on the GMI dataset

	Gene number	Mean 4.Ne.s	4.Ne.s credible interval	Conclusion
All genes	15,720	4.Ne.sgbM = 0.199	0.052–0.378	No selection on methylation (4.Ne.s ≪ 1)
Ancestrally gbM genes	1,383	4.Ne.sgbM = 0.866	0.368–1.341	Nonsignificant selection on gbM state (4.Ne.sgbM ≈ 1)
Ancestrally UM genes	6,078	4.Ne.sUM = 3.751	3.274–4.242	UM state advantageous (4.Ne.sUM ≫ 1)

Parameters of the SFS model were estimated using an mcmc approach (see Materials and Methods for details). The estimated mean selection efficacy (either 4.N_e.s_UM or 4.N_e.s_gbM, depending on which is higher) is shown with its 95% credible interval. If 4.N_e.s > 1.0 and if 1.0 is not included in the credible interval, then significant selection is inferred to act on methylation state (the UM or the gbM state is advantageous). Details on inferred values of other parameters of the model can be found in Supplementary Table S3. These results come from one mcmc run and are equivalent to results obtained from two independent runs with random parameter initiation values (Supplementary Table S3).