Table 1.
Site Models for Testing Positive Selection Affecting Amino Acid Residues in a Protein.
| Site model | Free parametersa | Number of site classes | Model parameters | Model comparison (LRT) |
|---|---|---|---|---|
| M1a | 2 (p0, ω0) | 2 |
p
0 (p1 = 1 − p0), ω0 < 1, ω1 = 1 |
M1a vs. M2a: df = 2 |
| M2a | 4 (p0, p1, ω0, ω2) | 3 |
p
0, p1 (p2 = 1 − p0 − p1), ω0 < 1, ω1 = 1, ω2 > 1 |
|
| M7 | 2 (p, q) | 10 | beta(p, q) | M7 vs. M8: df = 2 |
| M8 | 4 (p0, p, q, ωs) | 11 |
p
0, (p1 = 1 − p0), beta(p, q), ωs > 1 |
a
Number of free parameters in the ω distribution. M1a: p0 is the proportion of sites with ω0 < 1, while p1 =1 − p0 is the proportion of sites with ω1 = 1. M2a: same as M1a but includes an additional class of sites with ω2 > 1 in proportion p2, with p0 + p1 + p2 = 1. M7: the model uses a beta distribution with parameters p and q to describe variable ω for sites in the range 0 ≤ ω ≤ 1. M8: p0 is the proportion of sites with ω from beta(p, q) as in M7, but an additional class is now added (with proportion p1) with ωs > 1.