Chemical Similarity, Shape Matching and QSAR
E. V. RADCHENKO*ab, G. F. MAKHAEVAb, V. A. PALYULINab AND N. S. ZEFIROV ab
aLomonosov Moscow State university, department of Chemistry,
Lenkinskie gory 1/3, Moscow, 119991, Russia; binstitute of physiologically Active Compounds RAS, Severniy proezd 1, Chernogolovka, Moscow region, 142432, Russia
one of the fundamental paradigms of chemistry is the similarity property principle: “similar structures possess similar properties”.1,z it follows naturally from the chemical structure theory (structure determines the properties of a compound)^4 and the intuitive expectation that such structure-to-property functions are continuous (‘nature makes no leaps’). this principle lays the basis for the detection, analysis, and interpretation of patterns in the known data on the properties (including biological activities) of the compounds as well as for using these patterns to predict the properties for novel structures or to design the structures with desired properties. Although, as always, the devil is in the detail, and the specific applications of these approaches often have their own prerequisites and limitations (as we will see later), they
Issues in Toxicology No. 31
Computational Systems Pharmacology and Toxicology Edited by Dale E. Johnson and Rudy J. Richardson © The Royal Society of Chemistry 2017 Published by the Royal Society of Chemistry, www.rsc.org provide extremely useful tools for many areas of chemistry, pharmacology, and toxicology.
In addition to the purely scientific applications, the question of whether a given compound is substantially (or sufficiently) similar to some other compound may also have legal implications, in particular in the intellectual property and drug enforcement contexts. unfortunately, the criteria for solving it set forth in various jurisdictions and laws have significant differences and often tend to be too vague, too over-reaching and/or too restrictive. To some extent, this is caused by the attempts to prescribe simple solutions for complex structural problems that are, in turn, only indirectly related to the critical issue of whether a given compound will be substantially (or sufficiently) similar in properties to some other compound.
Due to the limited size of this chapter, we cannot hope to present a complete picture of all the methods developed over more than five decades in the fields of quantitative structure-activity relationships (QSAR)/quantitative structure-property relationships (QSPR) and molecular similarity analysis. Instead, the choice of the topics covered in the chapter will be a little subjective, with some focus on basic ideas and representative examples, on more recent results and on the techniques and services that are immediately available for solving some of the practical problems of computational pharmacology and toxicology. Among many excellent texts and reviews looking at somewhat different aspects of these areas, several publications^5-12 deserve a particular mention. The references to more detailed reviews on specific topics are also provided throughout the chapter when appropriate.
In the following sections, we first consider the concept of molecular similarity as a measure of closeness in chemical space and some approaches to the analysis of activity landscapes defined in such chemical spaces (Section 6.2). Then the approaches to the analysis of structure-activity/property relationships in congeneric and/or compact series of compounds that can be expected to share a consistent mechanism of action are discussed (Section 6.3.1). Finally, the approaches to the analysis of diverse and/or large data sets are presented (Section 6.3.2). This distinction reflects significant differences in the goals and methods of the analysis in these two situations.