If the three dimensional structure of one of the proteins to be aligned is known, then this information may be encoded in the form of a modified gap penalty [43][38][37]. The penalty reduces the likelihood of insertions/deletions occurring in known secondary structure regions, or conversely increases the likelihood of placing gaps in known loop regions. This approach increases the usual accuracy of alignment and has the additional bonus of reducing the sensitivity of the alignments to changes in gap penalty [37].
A stricter constraint on the alignment is possible if specific residues are known to be equivalent in the two proteins. The weight for aligning these specific residues may be increased to force them to align. However, if this type of treatment is really necessary, then it is likely that the alignment will have a low significance score and must be treated with caution.