Table 3.
Effect of core matrix type on accuracy of conditional model (TRCO) in 5-fold cross-validation experiments
| Model | acc(0) (%) | acc(1) (%) | Δ0,1 (%) | parms(0) | parms(1) | Y↔R | EQ |
|---|---|---|---|---|---|---|---|
| JC | 67.8 | 68.3 | 0.5 | 1 | 4 | no | no |
| FEL | 69.1 | 69.5 | 0.4 | 1 | 4 | no | yes |
| K2P | 73.6 | 74.9 | 1.3 | 2 | 8 | yes | no |
| HKY | 73.8 | 76.5 | 2.7 | 2 | 8 | yes | yes |
| GTR | 74.8 | 77.8 | 3.0 | 6 | 24 | yes | yes |
JC: Jukes and Cantor's (1969) model; K2P: Kimura's (1980) 2-parameter model; acc(n): classification accuracy for n-th order model; Δ0,1: acc(1)-acc(0); parms(n): number of free parameters (excluding equilibrium frequencies) in n-th order matrix; Y↔R: whether transition and transversion rates can be separately parameterized; EQ: whether non-uniform equilibrium frequences can be modeled.