Table 2. A comparison of codon usage between regions of HA1: epitopes A-E and the remaining residues, HA^-.

	Miyata	Hamming
Ep A vs. HA-	0.927	1.000**
Ep B vs. HA-	0.852	0.962
Ep C vs. HA-	0.917	0.998**
Ep D vs. HA-	0.974	1.000**
Ep E vs. HA-	0.816	0.982*
HA+ vs. HA-	0.999**	1.000**
P vs. HA-	0.993*	1.000**
HA- vs. HA-	0.506	0.506
HA+ vs. HA+	0.498	0.496

We also compare codon usage in the 18 residues of positive selection, denoted by P, identified by Bush et al. (15). Asterisks indicate statistical significance, as in Table 1. Note that, in general, epitopes are significantly biased relative to nonepitopes in the Hamming but not the Miyata metric. As a consistency check for our bootstrap methodology, we also compare HA⁺ and HA^- to themselves and find they lie within two standard deviations of the 50th percentile, as expected.