Skip to main content

Table 5 Comparison of various scoring matrices and scoring algorithms. The first column shows the scoring matrices. For example, BLOSUM62SWOPT is the matrix optimized for the SW algorithm starting from the BLOSUM62 matrix. The second column shows the performance of each score matrix by the SW algorithm on the COG distant test set. The following columns show the performance, in terms of average C, of each matrix used in combination with either the SW algorithm or the LA kernel on four different datasets. The best ⟨C⟩ in each column is highlighted in bold font.

From: Optimizing amino acid substitution matrices with a local alignment kernel

⟨C⟩

Score matrix

COG distant

COG close

PFAM distant

PFAM close

 

SW

LA

SW

LA

SW

LA

SW

LA

BLOSUM62

0.35

0.42

0.72

0.73

0.45

0.49

0.76

0.78

BLOSUM62SWOPT

0.42

0.42

0.70

0.71

0.49

0.51

0.75

0.77

BLOSUM62LAOPT

0.43

0.51

0.67

0.69

0.47

0.54

0.72

0.75

PAM250

0.36

0.35

0.53

0.52

0.43

0.43

0.62

0.61

PAM250SWOPT

0.38

0.37

0.57

0.56

0.44

0.44

0.65

0.64

PAM250LAOPT

0.26

0.36

0.53

0.46

0.43

0.38

0.62

0.56

GCB

0.13

0.12

0.33

0.32

0.12

0.11

0.37

0.36

GCBSWOPT

0.25

0.24

0.50

0.50

0.17

0.19

0.55

0.56

GCBLAOPT

0.14

0.15

0.39

0.31

0.16

0.093

0.44

0.38

JTT

0.34

0.34

0.53

0.51

0.47

0.46

0.61

0.60

JTTSWOPT

0.31

0.32

0.54

0.53

0.48

0.47

0.62

0.61

JTTLAOPT

0.34

0.31

0.53

0.45

0.48

0.38

0.61

0.55