Table 1. Results from maximum likelihood codon-based models of selection using Codeml.
| Model | P | ln L | Parameter estimates | Positively selected sites |
|---|---|---|---|---|
| M0 (one ratio) | 1 | -537.57 | K = 1.51, ω = 1.174 | None |
| M1a (nearly neutral) | 2 | -527.69 | K = 1.126, p 0 = 0.678, p 1 = 0.322, ω0 = 0, ω1 = 1 | Not allowed |
| M2a (positive selection) | 4 | -516.17 | K = 1.569, p 0 = 0.668, p 1 = 0.185, p 2 = 0.147, ω0 = 0, ω1 = 1, ω2 = 8.99 | 12Y, 39D, 56N, 60Y, 68G |
| M3 (discrete) | 5 | -516.13 | K = 1.627, p0 = 0.795, p1 = 0.186, p2 = 0.019, ω0 = 0.043, ω1 = 6.268, ω2 = 19.897 | 10D, 12Y, 13F, 29Y, 39D, 56N, 60Y, 68G |
| M7 (beta) | 2 | -529.17 | K = 1.23, p = 0.005, q = 0.005 | Not allowed |
| M8 (beta and omega) | 4 | -516.17 | K = 1.567, p 0 = 0.849, p 1 = 0.150, p = 0.0277, q = 0.0277, ω = 8.813 | 12Y, 39D, 56N, 60Y, 68G |
P = number of parameters in the ω distribution; K = estimated transition/transversion rate; ω = selection parameter; pn = proportion of sites that fall into the ωn site class; p, q = shape parameters of the β function (for models M7 and M8). Positively selected sites denoted in cursives were significant at P > 95%, while bold sites were significant at P > 99%.