Skip to main content

Table 3 Comparison of different structure alignment methods for three structure classes according to CATH [46]

From: Connectivity independent protein-structure alignment: a hierarchical approach

Program/Server

Mainly-α 19 str.a

Mainly-β 19 str.a

Mixed-α-β 15 str.a

Average performance (%)

CE

17

19

13

93

DALI

14

19

14

89

DEJAVU

14

19

9

79

GANGSTA

18 b

19

15

98

LOCK

0

14

11

47

MATRAS

11

19

14

83

SSM

5

13

10

53

TOP

14

18

12

83

TOPS

2

15

14

59

TOPSCAN

15

12

9

68

VAST

12

17

15

83

YAKUSA

17

19

14

94

  1. Except for GANGSTA all data were taken from literature [52, 53]. Average performances differ slightly, since structures with low secondary structure content were omitted. The 53 proteins of the Novotny dataset (see method section) were aligned against the SCOP40 reference database. For the GANGSTA evaluation the assignment of a reference structure was successful, if at least one target with appropriate CATH topology was found among the top 100 ranked protein domains.
  2. a Target protein structures belonging to the specified CATH class that are used for assignment to the appropriate CATH topology (for details see text).
  3. b Since protein 1c3u was moved to another topology class in more recent CATH versions, 18 is the maximum number of correct structure alignments achievable. Thus, preventing GANGSTA from reaching the maximum performance of 100% (Table S5 [see Additional file 1]).