Alignment Algorithm

We first define local neighbours of residues. Note that if we use all points for all possible triplets, each atom will have a redundant representation. It will appear in the hash table in all possible reference frames. (In practise, since we are not interested in very closed nor very distant atoms, we pick atoms in an annulus defined by min and max radii).
The Geometric Hashing technique is applied next using only neighboring points to detect seed matches defined by a transformation and a match-list.
Many of the matches obtained before represent the same transformation, i.e. different match lists may share the same transformation. We cluster seed matches and merge match-lists that were found.
The last step of the algorithm is the extending step. The seed matches are extended to contain additional matching pairs and best RMSD transformation is detected. For this purpose a heuristic iterative matching algorithm which minimizes the sum of the distances between the newly matched pairs is applied.