. 2006 Dec 5;7:529. doi: 10.1186/1471-2105-7-529

Table 1.

Coefficient of the approximation of σ²( $\hat{Δ}$ _pairwise)

Type	$\hat{d}$ _XY+ $\hat{d}$ _XZ	σ²( $\hat{d}$ _XY) + σ²( $\hat{d}$ _XZ)	${\hat{d}}_{Y Z}^{2}$	σ²( $\hat{d}$ _YZ)	σ²( $\hat{d}$ _XY) σ²( $\hat{d}$ _XZ)	error	dim
Day	-1.3090	1.0435	0.6895	-0.3339	0.1590	0.087	2.13
DNA	-1.2449	1.0933	0.6591	-0.3026	0.1181	0.098	2.13
JTT	-1.2921	1.0978	0.6741	-0.3065	0.1144	0.080	2.10

Coefficients of the regression on the logarithms for the three types of scoring matrices. The error column shows the mean error, which by virtue of being a regression on logarithms is very close to the relative error.