To build a model, all comparative modeling programs depend on a list of assumed structural equivalences between the target and template residues. This list is defined by the alignment of the target and template sequences. Although many template search methods will produce such an alignment, it is usually not the optimal target-template alignment in the more difficult alignment cases (e.g., at less than 30% sequence identity). Search methods tend to be tuned for detection of remote relationships, not for optimal alignment. Therefore, once the templates are selected, an alignment method should be used to align them with the target sequence. The alignment is relatively simple to obtain when the target-template sequence identity is above 40%. In most such cases, an accurate alignment can be obtained automatically using standard sequence-sequence alignment methods. If the target-template sequence identity is lower than 40%, the alignment generally has gaps and needs manual intervention to minimize the number of misaligned residues. In these low sequence identity cases, the alignment accuracy is the most important factor affecting the quality of the resulting model. Alignments can be improved by including structural information from the template. For example, gaps should be avoided in secondary structure elements, in buried regions, or between two residues that are far in space. Some alignment methods take such criteria into account [47,48,11,49,50]. It is important to inspect and edit the alignment in view of the template structure, especially if the target-template sequence identity is low. A misalignment by only one residue position will result in an error of approximately 4 in the model because the current modeling methods generally cannot recover from errors in the alignment.
When multiple templates are selected, a good strategy is to superpose them with each other first, to obtain a multiple structure-based alignment. In the next step, the target sequence is aligned with this multiple structure-based alignment. Another improvement is to calculate the target and template sequence profiles, by aligning them with all sequences from a non-redundant sequence database that are sufficiently similar to the target and template sequences, respectively, so that they can be aligned without significant errors (e.g., better than 40% sequence identity). The final target-template alignment is then obtained by aligning the two profiles, not the template and target sequences alone. The use of multiple structures and multiple sequences benefits from the evolutionary and structural information about the templates as well as evolutionary information about the target sequence, and often produces a better alignment for modeling than the pairwise sequence alignment methods [51,52].