Table 1. Results of PPA for the sets AE21 and AH21.
For each set, the results from nonrecoded and Dayhoff-6, Dayhoff-9, and Dayhoff-15 matrices are shown for comparison. The best fitting parameter in each group is highlighted in bold. In all cases CAT + GTR + G model was used. None of the recoding schemes has the absolute advantage in all five parameters.
| Matrix/recoding scheme | Model |
Site-specific amino
acid preferences |
Across-taxa
compositional heterogeneity |
Empirical
convergence probability |
Across-site
compositional heterogeneity |
Mean squared
heterogeneity across taxa |
| PPA-DIV | PPA-MAX | PPA-CONV | PPA-VAR | PPA-MEAN | ||
| Set AE21: | ||||||
| None | CAT + GTR | 6.50338 | 138.335 | 8.1909 | 8.8697 | 313.752 |
| Dayhoff-6 | CAT + GTR | −1.7959 | 82.4557 | 10.9337 | 11.0215 | 106.943 |
| Dayhoff-9 | CAT + GTR | −0.6288 | 45.4654 | 11.6434 | 12.1904 | 101.149 |
| Dayhoff-15 | CAT + GTR | 4.3768 | 98.1346 | 14.1153 | 13.8091 | 226.055 |
| Set AH21: | ||||||
| None | CAT + GTR | 4.9079 | 39.0857 | 2.83906 | 2.78946 | 48.7904 |
| Dayhoff-6 | CAT + GTR | 0.1449 | 18.7388 | 5.4639 | 5.8251 | 19.8551 |
| Dayhoff-9 | CAT + GTR | 0.8086 | 18.5237 | 6.0536 | 6.0197 | 20.1687 |
| Dayhoff-15 | CAT + GTR | 3.3539 | 33.3942 | 5.7612 | 4.9709 | 34.7670 |