Table 2

SS algorithm assignment success
Linkage metric m s 1-domain 2-domain 3-domain 4-domain
Average midpt 19 5 71% (0.52) 60% (0.35) 47% (0.28) 34% (0.19)
Average midpt 22 5 75% (0.55) 60% (0.38) 46% (0.31) 34% (0.21)
Average midpt 25 5 80% (0.56) 58% (0.40) 43% (0.32) 28% (0.20)
Average midpt 22 3 68% (0.53) 51% (0.29) 41% (0.20) 35% (0.15)
Average midpt 22 5 75% (0.55) 60% (0.38) 46% (0.31) 34% (0.21)
Average midpt 22 7 84% (0.53) 55% (0.40) 38% (0.33) 18% (0.18)
Average closest 22 3 79% (0.55) 51% (0.51) 40% (0.41) 27% (0.27)
Average closest 22 4 81% (0.54) 52% (0.38) 39% (0.28) 28% (0.19)
Complete midpt 40 5 71% (0.51) 45% (0.25) 38% (0.20) 22% (0.10)
Complete midpt 40 7 73% (0.51) 48% (0.28) 38% (0.23) 23% (0.13)
Complete midpt 40 9 77% (0.50) 50% (0.32) 38% (0.26) 18% (0.12)
Complete midpt 36 7 67% (0.49) 47% (0.24) 37% (0.19) 24% (0.11)
Complete midpt 38 7 70% (0.50) 47% (0.26) 38% (0.21) 23% (0.11)
Complete midpt 42 7 76% (0.51) 49% (0.31) 38% (0.25) 23% (0.14)

Given on the ASTRAL30 data set as a function of m and s. Linkage refers to the clustering technique used in determining the domains. Metric is either midpt, meaning distances between secondary structure elements were taken between their midpoints, or closest, meaning the closest approach distance was used. The optimal combination of m and s are shown in bold for each section of the table. Matthews correlation coefficient is given in parentheses.

Feldman

Feldman BMC Bioinformatics 2012 13:286   doi:10.1186/1471-2105-13-286

Open Data