Table 4.
Time complexities of different SVM-based methods for remote homology detection
| Methods | Time complexities |
| SVM-Top-n-gram-combine | O (n2l) + O (nml) + O (n2m) |
| SVM-Bprofile | O (n2l) + O (nml) + O (n2m) |
| SVM-Ngram | O (nl) + O (nml) + O (n2m) |
| SVM-Pattern | O (nllognl+n2l2m) + O (nml) + O (n2m) |
| SVM-Motif | O (n2l2) + O (nml) + O (n2m) |
| SVM-Top-n-gram-combine-LSA | O (n2l) + O (nml) + O (nmt) + O (n2R) |
| SVM-Bprofile-LSA | O (n2l) + O (nml) + O (nmt) + O (n2R) |
| SVM-Ngram-LSA | O (nl) + O (nml) + O (nmt) + O (n2R) |
| SVM-Pattern-LSA | O (nllognl+n2l2m) + O (nml) + O (nmt) + O (n2R) |
| SVM-Motif-LSA | O (n2l2) + O (nml) + O (nmt) + O (n2R) |
| SVM-Pairwise | O (n2l2) + O (n3) |
| SVM-LA | O (n2l2) + O (n3) |
| Profile | O (n2l) + O (n2l) |
| SW-PSSM | O (n2l) + O (n2l2) |
Where n is the number of training samples, l is the length of the longest training sequence, m is the total number of the words of each building block, t is the minimum of n and m and R is the length of the latent semantic feature vector.