Table 1. Likelihood ratio tests for positive selection among rodent and rabbit Trim5 homologs.

(A) Maximum likelihood trees of DNA sequences for RBCC (excluding exon8) and PRYSPRY (exon 8) are used for Trim5 paralogs from mice, rats and for rabbit Trim5α (see figure 4 for tree). The analysis for RBCC domains was repeated by excluding the Trim with the highest branch dN/dS value (Mmusc Trim12, see figure 4A). Similarly, the analysis for the PRYSPRY domain was repeated by excluding the Trim with the highest branch dN/dS value (Mdunni Tim30, see figure 4B) and also excluding rabbit Trim5 α.

(B) Mus species phylogeny (Tucker 2006) is used for Trim30 orthologs from mice and rats (see figure 5A for PRYSPRY tree). The analysis for exon8 domains was repeated by excluding Mus dunni Trim30 (the highest branch dN/dS value).

A RBCC (no exon8) of Trim5 paralogs from Mus musculus, Mus dunni, Rat, Rabbit.
	Codon model1	M7 vs M82	Tree length3
	f3×4	p<0.001	3.85
	f61	p<0.001	4.1
Excluding Mus musculus Trim12:	f61	p<0.001	4.06
PRYSPRY (exon8) of Trim5 paralogs from Mus musculus, Mus dunni, Rat, Rabbit.
	Codon model	M7 vs M8	Tree length
	f3×4	p=0.0021	4.39
	f61	p=0.006	4.42
Excluding Mus dunni Trim30:	f61	p=0.019	4.38
Excluding Rabbit Trim5:	f61	p=0.019	4.38
B SPRY (exon8) of Trim30 orthologs from mouse species and rat.
	Codon model	M7 vs M8	Tree length
	f3×4	p=0.003	1.63
	f61	p=0.0075	1.63
Excluding Mus dunni Trim30:	f61	p=0.014	1.52

¹Two different codon models are used (f3×4 and f61) to ensure that codon frequency models yield similar results.

²Likelihood ratio tests were done using log likelihood values obtained from PAML, comparing a null model (M7) to a positive selection model (M8). p values were obtained running a chi square test of log likelihood values, with 2 degrees of freedom.

³Tree length values obtained by PAML based on the input tree, used to assess evolutionary depth of sample input.