Table 3 Parameter estimates and likelihood values of the various models of codon evolution using CODEML in PAML
ModelModel Parameters-lnLModels Comparison2ΔLPr.Sites Under Positive Selection (Bayes Empirical Bayes, Pr. ω>1)
A. Test on the truncated and full-length MAT1-1-1 genes within a species (datasets with positive selection detected)
G. clavigera (9 isolates)M1a (neutral)P0 = 0.26, P1 = 0,742172.75M1a vs. M2a17.70<0.0012 G (0.99**), 3 I (0.99**), 8 N (0.99**)
M2a (selection)P0 = 0.99, P1 = 0.00, P2 = 0.004, ω2 = 227.052163.90
M7 (beta)P = 2.30, q = 0.0052172.75M7 vs. M817.41<0.0012 G (0.99**), 3 I (0.99**), 8 N (0.99**)
M8 (beta + ω)P0 = 0.98, P = 0.005, q = 3.27, P1 = 0.02, ω = 81.422164.05
G. aurea (four isolates)M1a (neutral)P0 = 0.77, P1 = 0.232102.86M1a vs. M2a21.18<0.0013 A (0.94), 8 A (0.94)
M2a (selection)P0 = 0.99, P1 = 0.00, P2 = 0.003, ω2 = 282.782092.27
M7 (beta)P = 0.005, q = 0,0122102.96M7 vs. M821.37<0.0013 A (0.97*), 8 A (0.97*)
M8 (beta + ω)P0 = 0.99, P = 0.005, q = 15.47, P1 = 0.003, ω = 282.822092.27
L. longiclavatum (five isolates)M1a (neutral)P0 = 0.72, P1 = 0.282183.24M1a vs. M2a18.88<0.0013 A (0.93), 8 A (0,93)
M2a (selection)P0 = 0.99, P1 = 0.00, P2 = 0.002, ω2= 209.512173.80
M7 (beta)P = 0.005, q= 0.0112183.25M7 vs. M818.90<0.0013 A (0.97*), 8 A (0.97*)
M8 (beta + ω)P0 = 0.99, P = 0.51, q = 1.49, P1 = 0.001, ω = 209.582173.80
B. Test on the truncated and full-length MAT1-1-1 genes within a species (dataset without positive selection detected)
L. terebrantis (4 isolates)M1a (neutral)2462.37M1a vs. M2a1.30ns
M2a (selection)2461.72
M7 (beta)2462.54M7 vs. M81.65ns
M8 (beta + ω)2461.72
  • Parameter estimates and likelihood values of the various models of codon evolution using CODEML in PAML. Notes for G. clavigera: Models assuming positive selection (M2a and M8) fit better the data than neutral models (M1a and M7) according to likelihood ratio tests. Model M2 assumes that 0.4% of the sites have dN/dS value = 227.05. Three sites under strong positive selection (P > 0.99) are identified by this model with Bayes empirical Bayes methods. Model M8 showed that approximately 98% of sites have dN/dS from a U-shaped beta distribution (hence, data fit strongly this model) and approximately 2% of site are under strong positive selection with dN/dS = 81.4. Both models M2 and M8 identified the same positive selection sites with Bayes empirical Bayes methods (even if model M8 would assume more sites under positive selection).