Current Topics in Medicinal Chemistry - Volume 6, Issue 1, 2006

The chemical universe containing organic molecules within a reasonable molecular weight is vast and largely unexplored. Estimations of possible numbers of unique molecules range from 10 13 to 101 80. These numbers have to be compared with the few tens of millions of compounds currently known. Design of libraries that populate the medicinally relevant chemical subspace and tools that help to maximise the chance of identifying leads are necessary. This review describes various molecular representations that lead to the definition of chemical space, drug space or activity space. Strategies for compound selection in such spaces are discussed, as well as potential sources of diversity that could be used to explore the medicinal space in quest of new drugs.

For many years, the most frequently optimized qualities of a screening library, or corporate compound collection, were size and diversity. Maximizing the number of diverse hits is the fundamental goal of such strategies. The ostensible justification that "bigger is better" is based on the large, estimated size of small-molecule space and the hypothesis that the notoriously low hit rates from high-throughput screening (HTS) could be overcome by brute force: i.e. by screening more compounds. Published, detailed studies about the success (or failure) of the brute-force strategy are rare, but it is well-known that it did not fulfill expectations. As a result, published reports in recent years have increasingly described methods for designing, selecting or synthesizing gene family-focused or -biased libraries. Moreover, many of the larger compound suppliers now sell such libraries, reflecting the growing interest in them from both the pharmaceutical and biotechnology markets. The trend towards gene family-focused libraries marks the emergence of a different hypothesis about how to increase HTS hit rates and also reflects an increasingly pragmatic focus on the management of screening libraries. An important, underlying assumption in this trend is that a high-quality, generalpurpose screening library of manageable size is neither realizable nor desirable. Whether a biasing strategy based on a specific gene family will do a better job of meeting both the scientific and business needs of the drug discovery enterprise still remains to be seen, but it is certainly an active area of current research. This review focuses on the "who, what, why, when, and how" of the design of gene family-focused libraries. Particular attention is given to reports that discuss not only the techniques used, but also any results obtained.

Decision trees are among the most popular of the new statistical learning methods being used in the pharmaceutical industry for predicting quantitative structure-activity relationships. This article reviews applications of decision trees in drug discovery research and extensions to the basic algorithm using hybrid or ensemble methods that improve prediction accuracy.

To be effective, a designed drug must discriminate successfully the macromolecular target from alternative structures present in the organism. The last few years have witnessed the emergence of different computational tools aimed to the understanding and modeling of this process at molecular level. Although still rudimentary, these methods are shaping a coherent approach to help in the design of molecules with high affinity and specificity, both in lead discovery and in lead optimization. It is the purpose of this review to illustrate the array of computational tools available to consider selectivity in the design process, to summarize the most relevant applications, and to sketch the challenges ahead.

Molecular modeling efforts aimed at probing the structure, function and inhibition of the acetylcholinesterase enzyme have abounded in the last decade, largely because of the system's importance to medical conditions such as myasthenia gravis, Alzheimer's disease and Parkinson's disease, and well as its famous toxicological susceptibility to nerve agents. The complexity inherent in such a system with multiple complementary binding sites, critical dynamic effects and intricate mechanisms for enzymatic function and covalent inhibition, has led to an impressively diverse selection of simulation techniques being applied to the system, including quantum chemical mechanistic studies, molecular docking prediction of noncovalent complexes and their associated binding free energies, molecular dynamics conformational analysis and transport kinetics prediction, and quantitative structure activity relationship modeling to tie salient details together into a coherent predictive tool. Effective drug and prophylaxis design strategies for a complex target like this requires some understanding and appreciation for all of the above methods, thus it makes an excellent case study for multi-tiered pharmaceutical modeling. This paper reviews a sample of the more important studies on acetylcholinesterase and helps to elucidate their interdependencies. Potential future directions are introduced based on the special methodological needs of the acetylcholinesterase system and on emerging trends in molecular modeling.