Molecular Similarity, Chemical Spaces and Activity Landscapes
Molecular Similarity: Concept and Definitions
Molecular Similarity Concept
The idea that some molecules are more or less similar to some other molecules may seem obvious to anyone who has ever studied chemistry. Indeed, it was recognized in the earliest years of chemical science5 and forms an important framework for the systematic analysis and presentation of the accumulated chemical knowledge. However, as is often the case with
‘intuitively obvious’ concepts, the concept of molecular similarity has many aspects, variations, definitions, and applications, as evidenced by the vast body of literature on the subject.^5 In most general terms, we can consider three components of a similarity measure: applicability domain, structure representation, and similarity (comparison) function. These components are not independent but more or less closely related to each other, and a choice for one of them usually limits the possible (or desirable) choices for others. Nevertheless, within these limits one can combine different approaches almost as freely as the units of a model railway. This opens immense possibilities for the molecular similarity analysis but also greatly complicates the field in practice.