Skip to main content

Table 2 Model comparison for full MTBC dataset

From: A seventeenth-century Mycobacterium tuberculosis genome supports a Neolithic emergence of the Mycobacterium tuberculosis complex

Model Marginal likelihood Mean rate (95% HPD) Mean rate variance (95% HPD) Mean tree height (95% HPD)
BDSKY+UCLD − 6125044.47176458 1.4488E−8 (9.4606E−9, 1.9632E−8) 1.6881E−17 (5.4855E−18, 3.069E−17) 3258.0478 (2189.5235, 4501.1384)
CC+UCLD − 6126017.15694528 1.214E−8 (7.1934E−9, 1.6448E−8) 1.2459E−17 (2.833E−18, 2.3969E−17) 4172.1961 (2585.2349, 6119.744)
SKY+UCLD − 6127733.35000634 1.2944E−8 (8.6149E−9, 1.7342E−8) 1.3423E−17 (4.848E−18, 2.3869E−17) 3650.4222 (2472.6434, 4992.0277)
CC+strict − 6125541.68118691 1.1573E−8 (8.6397E−9, 1.4509E−8) NA 4453.1162 (3330.1516, 5619.3974)
  1. Marginal likelihood and parameter estimates from four models applied to the full MTBC dataset: constant coalescent with uncorrelated lognormal clock (CC+UCLD), constant coalescent with strict clock (CC+strict), Bayesian skyline coalescent with uncorrelated lognormal clock (SKY+UCLD), and birth-death skyline with uncorrelated lognormal clock (BDSKY+UCLD). Marginal likelihoods obtained through path sampling (see the “Methods” section)