Table 1.

PAML-Implemented Site Models of Positive Selection

Modela	ln L	Parameter Estimatesb	Selected Codonsc
All species, full gene
M1a	−679.00
M2a	−673.02*	p0 = 0.436 (ω0 = 0.032), p1 = 0.529, p2 = 0.035 (ω2 = 21.88)	24 (0.82), 27 (0.99), 36 (0.88)
M7	−679.12
M8A	−678.99
M8	−673.06*	p0 = 0.44 (p = 0.02, q = 0.01), p2 = 0.035 (ω2 = 21.39)	24 (0.85), 27 (0.99), 36 (0.91)
All species, exon 3 excluded
M1a	−413.81	p0 = 0.876 (ω0 = 0.209), p1 = 0.124
M2a	−413.80
M7	−413.71	p = 0.758, q = 1.756
M8a	−413.71	p0 = 0.999 (p = 0.747, q = 1.725), p1 = 0.001
M8	−413.71
Human excluded, full gene
M1a	−619.31	p0 = 0.389 (ω0 = 0.0), p1 = 0.611
M2a	−618.26
M7	−619.32	p = 0.008, q = 0.005
M8a	−619.31	p0 = 0.389 (p = 0.005, q = 1.554), p1 = 0.611
M8	−618.26

*P < 0.05, comparing model including positive selection (M2a or M8) to nearly neutral model (M1a, M7, or M8a).

^aAll models and comparisons among models are described in the text. Asterisks indicate comparisons where models that incorporate a class of sites affected by positive selection are better fits to the data than nearly neutral models (Models M2a and M8 in the “all species, full gene” comparison); in all other comparisons, models incorporating positive selection are not better fits to the data than nearly neutral models.

^bParameter abbreviations are as follows: p₀ = proportion of sites falling into nearly neutral site class (followed by estimate of ω₀ for models M1a and M2a or shape parameters of the beta distribution (p,q) for models M7, M8a, and M8), p₁ = proportion of sites falling into neutral site class (ω₁ = 1); p₂ = proportion of sites falling into positively selected site class (followed by estimate of ω₂).

^cBEB estimates of sites falling into positively selected class are listed for those where the posterior probability (in parentheses) is greater than 0.8.