MOLECULAR SIMILARITY BASED ON NOVEL ATOM-TYPE ELECTROTOPOLOGICAL STATE INDEXES

A molecular similarity measure is developed from a structure represent ation based on atom level topological and electronic information, spec ifically the electrotopological state indices. For the purpose of crea ting a structure information space, atom typing is introduced. The cla ssification of each atom in the molecular graph is derived from its va lence state. The electrotopological state indices for all atoms of the same type in the molecule are summed to create an atom-type E-state i ndex for each atom type. These indices are shown to encode significant electronic and topological information. Further, they are shown to be very useful in the representation of structure information as a basis for molecular similarity judgment and for database characterization, including searching. Examples are given for similarity among a set of simple structures and for database searching using four drug molecules .