Skip to main content

Table 2 Pairwise comparisons between HSSP S and other matrices estimated from the FastMG procedure

From: FastMG: a simple, fast, and accurate maximum likelihood procedure to estimate amino acid replacement rate matrices from large data sets

M1

M2

LogLK (M1-M2)

M1 > M2 (#TP)

M1 < M2 (#TP)

#M1 > M2 (#TP)

#M1 < M2 (#TP)

HSSPS

HSSP 8 R

0.02

201 (130)

99 (65)

66 (27)

12 (4)

HSSPS

HSSP 16 R

0.01

208 (126)

92 (64)

69 (26)

10 (4)

HSSPS

HSSP 24 R

0.01

203 (114)

97 (72)

72 (26)

9 (5)

HSSPS

HSSP 32 R

0.01

206 (119)

94 (59)

69 (24)

11 (4)

HSSPS

HSSP 64 R

0.01

200 (101)

100 (63)

73 (16)

11 (4)

HSSPS

HSSP 8 T

0.01

191 (124)

109 (75)

43 (19)

33 (15)

HSSPS

HSSP 16 T

0.00

152 (95)

148 (81)

26 (5)

34 (9)

HSSPS

HSSP 24 T

0.00

142 (88)

158 (89)

19 (4)

36 (11)

HSSPS

HSSP 32 T

0.00

132 (78)

168 (87)

18 (3)

32 (6)

HSSPS

HSSP 64 T

0.00

131 (72)

169 (80)

15 (4)

40 (4)

  1. LogLK: the log likelihood difference per site between trees inferred using M1 and M2; a positive (negative) value means M1 is better (worse) than M2. M1 > M2: the number of alignments where M1 results in better likelihood value than M2; #TP: the number of alignments where tree topologies inferred using M1 and M2 are different. #M1 > M2 (p <0.05): the number of alignments where the Kishino-Hasegawa test indicates that M1 is significantly better than M2. #M1 < M2 (p <0.05): the same as #M1 > M2, but now M2 is significantly better than M1.