Fig. 2.

Simplified example of fingerprint algorithm. Two molecules are represented as collections of fitting points, shown as circles for one molecule, triangles for the other. The fitting points represent two types of chemical features, red and green. The molecules contain approximately congruent triangles of fitting points (outlined in black), the “core triplets”. The fitting points of each molecule are oriented so that their core triplets are approximately superimposed, and a grid placed over the resulting fitting-point assembly (bottom left). Each fitting point, except those of the core triplets, is mapped to the nearest grid point. For example, the red triangle at the top is mapped to the grid point shown by the purple arrow. A bit string is created for each molecule to represent grid-point occupancy, the first (last) 16 bits capturing occupancy by red (green) features. For example, the second bit for the triangle molecule is set to 1 because the second grid point (counting in rows starting at top left) is occupied by a red fitting point. There are two positions in which both bit strings have on bits, revealing the close proximity of the circled fitting points