Table 4.
Average Q scores (in %) on PREFAB 4.0
| TCoffee |
MUSCLE |
ProbCons |
MUMMALS |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Q(2) | ||||||||||||
| 0–20% {887} | 37.92 | 38.27 | 1 × 10−5 | 38.22 | 39.69 | 8 × 10−8 | 38.95 | 40.17 | 7 × 10−31 | 43.59 | 43.62 | 0.005 |
| 20–40% {588} | 82.60 | 82.92 | 4 × 10−8 | 81.75 | 83.87 | 1 × 10−29 | 82.84 | 84.30 | 4 × 10−39 | 85.39 | 85.45 | 2 × 10−4 |
| 40–70% {112} | 96.37 | 96.51 | 0.005 | 96.24 | 96.58 | 0.01 | 96.41 | 96.83 | 5 × 10−6 | 96.59 | 96.75 | 5 × 10−4 |
| 70–100% {95} | 97.94 | 98.04 | – | 97.97 | 97.91 | – | 97.76 | 98.05 | 3 × 10−4 | 97.75 | 97.93 | 0.03 |
| All {1682} | 60.82 | 61.13 | 1 × 10−12 | 60.68 | 62.21 | 7 × 10−29 | 61.44 | 62.64 | 3 × 10−71 | 64.79 | 64.85 | 5 × 10−8 |
| Q(50) | ||||||||||||
| 0–20% {887} | 49.67 | 50.00 | 6 × 10−6 | 50.71 | 50.95 | – | 55.63 | 55.72 | 0.02 | 57.68 | 57.91 | 0.02 |
| 20–40% {588} | 83.94 | 84.20 | 8 × 10−7 | 85.09 | 85.13 | – | 87.24 | 87.38 | 3 × 10−7 | 87.24 | 87.30 | 0.02 |
| 40–70% {112} | 95.99 | 95.55 | 0.02* | 94.72 | 96.46 | – | 95.39 | 95.48 | 0.004 | 95.34 | 95.41 | – |
| 70–100% {95} | 97.97 | 98.04 | – | 97.50 | 97.69 | – | 97.26 | 97.40 | 0.001 | 96.68 | 97.04 | 0.005 |
| All {1682} | 67.46 | 67.70 | 2 × 10−9 | 68.30 | 68.57 | – | 71.68 | 71.79 | 1 × 10−7 | 72.73 | 72.89 | 5 × 10−4 |
Each subset includes all structure pairs with protein sequence identity within the specified range, with * indicating worse accuracy in P-value. The Q(2) scores are obtained from aligning only the original input protein sequence pair, while the Q(50) scores are obtained from aligning the full set of protein sequences (at most 50) that also include random hits from database search and evaluations are made on the original input sequence pair. For each algorithm, the higher accuracy value is in bold.