Table 1. Parameter estimates and likelihood ratio statistics (2Δl) for the M. martensii α-toxin gene.
| Models | p | l | Parameter estimates | 2Δl | PSSs |
|---|---|---|---|---|---|
| M0 (one-ratio) | 1 | −1497.0 | ω = 0.85 | None | |
| M1a (neutral) | 2 | −1448.0 | p0 = 0.46 (p1 = 0.54) | Not allowed | |
| ω0 = 0.08 (ω1 = 1) | 16.6 | ||||
| M2a (selection) | 4 | −1439.7 | p0 = 0.42 | 2R, 8K*, 9P*, 15E, 17A, | |
| p1 = 0.31 (p2 = 0.27) | 18R**, 37Q, 38W, 39V**, | ||||
| ω0 = 0.09 (ω1 = 1) | 54N | ||||
| ω2 = 2.47 | |||||
| M7 (beta) | 2 | −1450.2 | p = 0.20, q = 0.15 | 20.0 | Not allowed |
| M8 (beta & ω) | 4 | −1440.2 | p0 = 0.64 (p1 = 0.36) | 2R*, 8K**, 9P**, 10H, 13V, | |
| p = 0.32, q = 0.58 | 15E*, 17A*, 18R**, 37Q*, | ||||
| ωs = 2.21 | 38W*, 39V**, 54N* |
Note: p is the number of parameters in the ω distribution; l is the log likelihood. Twice the log likelihood differences (2Δl) between null models and their alternative models: M1a/M2a = 16.6, M7/M8 = 20.0, both having χ2 significant values (p < 0.001). PSSs identified by the Bayes Empirical Bayes (BEB) methods under M2a and M8 with P (posterior probabilities) ≥ 0.9 are shown and those ≥ 0.99 are indicated by ** and ≥ 0.95 by *. The Naïve Empirical Bayes (NEB) methods gave similar results (Table S1). Residues are numbered according to BmKM1. Two ω values in M2a and M8, as indicators of positive selection, are boldfaced. M8-specific PSSs are underlined once.