Table 2.
Average SPS and CS scores (in %) on full length protein sequences in BAliBASE 3.0
| TCoffee |
MUSCLE |
ProbCons |
MUMMALS |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SPS | ||||||||||||
| 1V1 {38} | 53.81 | 54.21 | 56.21 | 56.98 | 64.46 | 64.48 | 64.41 | 64.23 | ||||
| 1V2 {44} | 91.55 | 91.98 | 90.62 | 91.50 | 93.50 | 93.65 | 93.53 | 94.00 | ||||
| 1 (V1–V2) {82} | 74.06 | 74.48 | 0.02 | 74.67 | 75.50 | 0.003 | 80.05 | 80.13 | – | 80.03 | 80.20 | – |
| 2 {41} | 89.04 | 88.82 | 88.08 | 88.24 | 89.93 | 89.94 | 89.18 | 89.39 | ||||
| 3 {30} | 71.09 | 71.19 | 75.01 | 76.27 | 78.62 | 78.30 | 80.76 | 80.79 | ||||
| 4 {49} | 82.21 | 82.37 | 84.83 | 85.64 | 87.43 | 87.25 | 83.69 | 83.97 | ||||
| 5 {16} | 81.94 | 80.98 | 82.69 | 82.83 | 87.69 | 87.87 | 86.33 | 87.40 | ||||
| All (1–5) {218} | 78.88 | 78.97 | 0.04 | 80.11 | 80.82 | 0.006 | 83.93 | 83.89 | – | 83.14 | 83.39 | – |
| CS | ||||||||||||
| 1V1 {38} | 31.34 | 32.21 | 35.63 | 33.95 | 40.45 | 41.00 | 41.61 | 41.39 | ||||
| 1V2 {44} | 81.64 | 82.68 | 80.75 | 82.93 | 85.52 | 85.77 | 83.98 | 86.41 | ||||
| 1 (V1–V2) {82} | 58.33 | 59.29 | 1×10−4 | 59.84 | 60.23 | 0.01 | 64.63 | 65.02 | 0.02 | 64.34 | 65.55 | – |
| 2 {41} | 37.85 | 38.88 | 35.27 | 37.61 | 40.63 | 40.49 | 42.83 | 43.46 | ||||
| 3 {30} | 36.00 | 36.83 | 40.57 | 42.73 | 54.37 | 54.80 | 49.40 | 49.57 | ||||
| 4 {49} | 48.20 | 48.78 | 47.37 | 49.67 | 53.67 | 53.14 | 48.55 | 49.76 | ||||
| 5 {16} | 50.63 | 49.31 | 47.94 | 44.94 | 57.38 | 57.31 | 52.88 | 57.00 | ||||
| All (1–5) {218} | 48.56 | 49.27 | 7×10−9 | 48.89 | 50.07 | 0.002 | 55.71 | 55.77 | 0.04 | 53.85 | 55.02 | 0.001 |
Reference 1 contains alignments of sequences that are subdivided into two subsets 1V1 (<20% identity) and 1V2 (20–40% identity). Reference 2 contains alignments that include orphan sequences. Reference 3 contains alignments of clusters of sequences from different families. Reference 4 contains alignments of sequences with large terminal extensions, while reference 5 contains alignments of sequences with internal insertions. The number in braces denotes the number of alignments in each subset. For each algorithm, the first number shows the accuracy of the original algorithm (TCoffee, MUSCLE, ProbCons, MUMMALS) that does not use horizontal information. The second number shows the accuracy of the modified algorithm NRAlign that makes use of horizontal information, with the higher accuracy value in bold. The third number shows the P-value, with – indicating insignificant differences. Since many of the subsets are small, P-values are computed only for reference 1 and for the entire set.