Table 2.

The linear models between evolutionary rates and amino acid compositions.

Gene name	Linear regression models (Ridge regression)		Linear regression models (Principal component regression)
	R2	Variables significantly contribute to evolutionary rates (positive; negative)	R2	Amino acid significantly contribute to the first Principal component (positive; negative)
alaS	0.93	F,H,T; Y,M,W,E,C,D,I,P	0.7	A,D,G,H,S,R; C,E,I,K,N,Y
ileS	0.93	A,G,Q; D,C,V,W	0.84	A,D,G,M,P,R,W; E,F,I,K,L,N,S,Y
argS	0.77	Q; S,E,I	0.69	A,D,G,H,M,Q,P,R,T; C,E,F,I,K,N,S
hisA	0.88	Y,L,R,A; D,M,E	0.77	A,D,G,H,L,Q,P,R,T; E,F,I,K,M,N,S
glnA	0.64	S,A,C,Q; D	0.46	A,G,H,P,S,T; C,E,F,K,L,N
purM	0.72	P,N; H,T,I	0.55	A,D,G,H,L,Q,P,R,T,W,V; E,F,I,K,M,N,Y

All multiple linear models in this table have P-values less than 1.539E-06. Amino acids coded by GC-rich codons are A, P, R, and G, while amino acids coded by AT-rich codons are F, I, N, K, M, and Y.