Table 4

The effects of codon usage bias on LRT (n = 12)



Model





F equal
F3x4
F codon

Estimates under model A

p0 = 0.51693, p1 = 0.152
(p2 + p3) = 0.330
ω2 = 11.182
p0 = 0.503, p1 = 0.102
(p2 + p3) = 0.393
ω2 = 14.201
p0 = 0.477, p1 = 0.095
(p2 + p3) = 0.426
ω2 = 999





Estimates under M1a

p0 = 0.732, p1 = 0.267
ω0 = 0.187
p0 = 0.793, p1 = 0.206
ω0 = 0.073
p0 = 0.794, p1 = 0.205
ω0 = 0.062





Estimates under model A (ω2 = 1)

p0 = 0.492, p1 = 0.159
(p2 + p3) = 0.348
ω0 = 0.165
p0 = 0.537, p1 = 0.112
(p2 + p3) = 0.350
ω0 = 0.073
p0 = 0.533, p1 = 0.112
(p2 + p3) = 0.354
ω0 = 0.039



2Δl
Test 1:
22.85
33.78
35.27

Test 2:
10.31
6.84
6.83

Note: Branch-site model A was compared to site model M1a and model A with ω2 fixed to 1 for test 1 and 2, respectively. All tests are significant at 1%; with df = 2 and df = 1 for tests 1 and 2, respectively.

Levasseur et al. BMC Evolutionary Biology 2006 6:92   doi:10.1186/1471-2148-6-92