Computational library design clustering

Property space Multidimensional representation of a set of compounds a.s poiiiLS in space. Each axis of the space represents some descriptor, either whole pniperiy or computed from the two-dimensional (2D) or three diinensional (3D) chemical stnicture. Compounds (hat aie similar to each other chemically will cluster together in propeny space. A further u.ssumption used in library design is that structurally similar compound.s will share similar biological activity. 

The concepts of molecular similarity (1-3) and molecular diversity (4,5) play important roles in modern approaches to computer-aided molecular design. Molecular similarity provides the simplest, and most widely used, method for virtual screening and underlies the use of clustering methods on chemical databases. Molecular diversity analysis provides a range of tools for exploring the extent to which a set of molecules spans structural space, and underlies many approaches to compound selection and to the design of combinatorial libraries. Many different similarity and diversity methods have been described in the literature, and new methods continue to appear. This raises the question of how one can compare different methods, so as to identify the most appropriate method(s) for some particular application this chapter provides an overview of the ways in which this can be carried out, illustrating such comparisons by,... 

Cluster-based and dissimilarity-based methods for compound selection were first discussed in the Eighties but it is only in the last few years that the area has attracted substantial attention as a result of the need to provide a rational basis for the design of combinatorial libraries. The four previous sections have provided an overview of the main types of selection method that are already available, with further approaches continuing to appear in the literature. Given this array of possible techniques, it is appropriate to consider ways in which the various methods can be evaluated, both in absolute terms and when compared with each other. A method can be evaluated in terms of its efficiency, /.< ., the computational costs associated with its use, and its effectiveness, /.< ., the extent to which it achieves its aims. As we shall see, it is not immediately obvious how effectiveness should be quantified and we shall thus consider the question of efficiency first, focusing upon the normal algorithmic criteria of CPU time and storage requirements. 

A primary, or unbiased, library is a large set of compounds (t5q)ically thousands to millions) based on diversity and aimed at the discovery of samples of interest for targets for which little, if any, information is available. Diversity is a concept unrelated to the library size that attempts to evaluate the representation of chemical space by a chemical library using computational methods If this space is sampled evenly by the components of a library, then this library is considered to be diversity based (Fig. 4.1, left). A focused, or biased, library is a similarity-based set of compounds (typically hundreds to thousands) aimed at the discovery and optimization of lead structures for a target for which a structural model on which to design the hbrary is available. Similarity is a concept unrelated to the library size that is opposite to diversity if the library components are clustered around the model structure A, the library is similarity based (see Fig. 4.1, right). 

---