Table 2. BAliBASE Q scores on subsets.
| Method | Ref1 | Ref2 | Ref3 | Ref4 | Ref5 |
|---|---|---|---|---|---|
| MUSCLE | 0.887 | 0.935 | 0.823 | 0.876 | 0.968 |
| MUSCLE-p | 0.871 | 0.928 | 0.813 | 0.857 | 0.974 |
| T-Coffee | 0.866 | 0.934 | 0.787 | 0.917 | 0.957 |
| NWNSI | 0.867 | 0.923 | 0.787 | 0.904 | 0.963 |
| CLUSTALW | 0.861 | 0.932 | 0.751 | 0.823 | 0.859 |
| FFTNS1 | 0.838 | 0.908 | 0.708 | 0.793 | 0.947 |
The average Q score for each method on each BAliBASE subset is shown. Ref1 is the largest subset with 81 test sets, comprising almost 60% of the database. Other subsets are smaller. For example, Ref4 and Ref5 have 12 alignments each, and there are large variances in the individual scores from which the averages are computed. In our opinion, it is not possible to draw meaningful conclusions about the relative performance of different methods on these subsets.