Considerable efforts within the scientific community are aimed at finding 3D pharmacophoric fingerprint descriptors. Unfortunately the results so far obtained show no clear improvement when compared with the 2D ones.

Our work in this field consists in properly cutting-off those pharmacophores that are less relevant for a given molecule and then compressing the relevant 3D information into a short 1024 bit string. We proposed to call these fingerprints "3D Mill hashed fingerprints". We develop these descriptors with the aim of finding molecules "pharmacophorically" similar but belonging to different chemical classes.

Other applications include the identification of lead follow-up molecules, monomer selection/replacement and small set creation for directed screening. These fingerprints proved to be very useful when combined with Daylight 2D hashed fingerprints.

Presentation slides: