Use of Biological Fingerprints Versus Structure/Chemotypes to Describe Molecules

Molecules are usually described by their chemical structure and by fingerprints derived from this. These range from 2D structure based, that only represent the underlying structure that gives rise to the properties recognized by a biological target to 3D pharmacophores or molecular interaction fields, that much better represent how the protein binding sites would “see” a molecule. However, all of these have many limitations, including conformation for the 3D structure-based approaches. More recently, experimental profiling data have been generated that enables a molecule to be described by a fingerprint of binding affinity to a diverse set of biological targets (pharmacological and “antitargets” such as CYP450 metabolic enzymes). These results show that small changes in structure can cause large changes in broad biological profile, and that a structure-based analysis/clustering of compounds, such as different hits, leads, or clinical candidates, often does not provide a differentiation that is relevant in biological space. The data show that “selective” versus “nonselective” compounds, and the type of off-target effects are not evident from a “chemotype” approach. The concept of “biological fingerprints” as a better way to describe compounds of biological interest is described in this chapter, focusing on the power of these descriptors versus structure-based descriptors to differentiate compounds and enable the selection of the best lead compounds. Keywords: biological profiling; chemotype; drug design; fingerprints; pharmacological profiling; pharmacophore; selectivity