A review of methods employed for the assessment of druglikeness using 2D
structural and atom type descriptors is presented. These methods are classified as Druglike
Filters (DLFs) and Druglike Indices (DLIs), depending on the characterization of
druglikeness, using known drug and non-drug databases. The DLF methods specify a
set of rules based on calculated property distributions, whereas the DLI methods aim to
assess druglikeness through a single number derived from multiple descriptors. A
review of ranges calculated from property profiles of known drugs is given, along with
a careful re-assessment for twenty five descriptors based on an analysis of a recent drug
database. A discussion of future direction for the development and utility of these
approaches is presented.
Keywords: Lead-likeness, drug likeness, structural descriptors, drug like index,
atom type diversity, relative drug-likeness potential, ALOGP, UALOGP, druglike
descriptors, chembridge database, drugs database, drug properties, atom
classification, structural diversity, lead optimization.
