Fast similarity measure for 3D structures.
This class implements a fast descriptor based 3D similarity measure described by Ballester et al
([Bellester, P.J. and Richards, W.G.
, Ultrafast shape recognition to search compound databases for similar molecular shapes, Journal of Computational Chemistry, 2007, 28:1711-1723]). The approach calculates the distances of each atom to four specific points: the
centroid of the molecule, the atom that is closest to the centroid, the atom that is farthest from the
centroid and the atom that is farthest from the previous atom. Thus we get 4 sets of distance distributions.
The final descriptor set is generated by evaluating the first three moments of each distance distribution.
The similarity between two molecules is then evaluated using the inverse of a normalized
Manhattan type metric.
This class allows you to evaluate the 3D similarity between two specified molecules as well as
generate the 12 descriptors used to characterize the 3D structure which can then be used for a
variety of purposes such as storing in a database.
Note: The methods of this class do not perform hydrogen removal. If you want to
do the calculations excluding hydrogens, you'll need to do it yourself. Also, if the molecule has
disconnected components, you should consider one (usually the largest), otherwise all components
are considered in the calculation.