This review gives an almost complete (its not a book, but 30 pages) overview about structure handling (SMILES, SDF), molecular databases (NCI,MDDR), molecular descriptors and of course similarity and dissimilarity calculations for small molecules. This is especially helpful for comparing large drug databases and calculating the overlap of these databases, or creating unique databases in silico with special designed properties. Different criterions are presented, ie. pharmaceutical or spectral criterions which lead to the selection of different descriptors (2D. 3D, Balaban index, logP, CoMFA descriptors etc) for dissimilarity calculations.

Drug companies always used "nature" for getting new inspirations, hence new ideas for molecules (diversity sets) and their usuage for treating diseases as a long term goal. However just comparing synthetic compound libraries with natural compound databases doing diversity calulations may not lead to direct usable ideas. Here the authors developed a hybrid library ‘MetaFocus’ (Metabolite Focused) library from Available Chemicals Directory (ACD) and the CRC Dictionary of Natural Products (DNP) molecules. "The strategy relies to a large extent on molecular similarity calculations to combine information from natural and synthetic molecules". Synthetic mimics of difficult to synthesize natural products can be used to built new lead molecules.

See also: Design of Array-Type Compound Libraries that Combine Information from Natural Products and Synthetic Molecules DOI - The Dictionary of Natural Products lists ~60.000 natural products (V code assigned, year 2006) which are parent structures (without additional modifications). See the Dictionary Subset of the Combinded Chemical Dictionary.