Table 2.
Comparison of alignment accuracy between ClustalW, MAFFT, SparseMap and mBed.
| Method | Alignment Column Score (%) | |
|---|---|---|
| BAliBase | HOMSTRAD/Pfam | |
| Guide Trees constructed internal to method | ||
| ClustalW | 32.66 | 60.12 |
| ClustalW -quicktree -ktuple=2 | 32.84 | 59.92 |
| MAFFT | 31.97 | 66.51 |
| MAFFT -retree 1 | 31.24 | 60.09 |
| MAFFT -retree 1 -parttree | 30.04 | 59.27 |
| MUSCLE | 35.80 | NA |
| MUSCLE -maxiters 1 | 32.04 | 60.45 |
| Guide Trees constructed external to method | ||
| MUSCLE+mBed | 35.38 | NA |
| MUSCLE -maxiters 1+ mBed | 32.86 | 64.18 |
| MAFFT + mBed | 29.79 | 57.57 |
| ClustalW + "MAFFT -retree 0 -parttree" | 31.64 | 54.75 |
| ClustalW + SeedMap | 29.82 | 58.85 |
| ClustalW + mBed | 30.20 | 59.24 |
| # of alignments | 386 | 646 |
Average Column scores (%) are given for each method. Accuracies are measured on two datasets. The HOMSTRAD-Pfam dataset comprises 646 test cases. Each test consists of a Pfam alignment containing between 3-10,000 sequences, which has been paired with a corresponding structural alignment from HOMSTRAD.