1. A root node is selected randomly among all atoms of
the structure except hydrogen. Structural information on
each node (atom or functional group) is converted into a
double-byte atom code in terms of elements, connectivity,
the number of single bonds, the size of the smallest ring,
the largest ring, charges, and so on. Although the code is
not a full representation of a connection table, it includes
ring information as a new matter. Thereby, the screening
process can finish in one stroke without extra procedures