Send Orders for Reprints to reprints@benthamscience.net Current Topics in Medicinal Chemistry, 2014, 14, 000-000 1 1568-0266/14 $58.00+.00 © 2014 Bentham Science Publishers Virtual Screening Strategies in Medicinal Chemistry: The State of the Art and Current Challenges Rodolpho C. Braga 1,2 , Vinícius M. Alves 1 , Arthur C. Silva 1 , Marilia N. Nascimento 1 , Flavia C. Silva 1 , Luciano M. Lião 2 and Carolina H. Andrade 1,* 1 LabMol - Laboratory for Molecular Modeling and Drug Design, Faculdade de Farmácia, Universidade Federal de Goiás, Rua 240, Qd.87, Setor Leste Universitário, Goiânia, Goiás, 74605-510, Brazil; 2 Laboratório de Ressonância Magnética Nuclear, Instituto de Química, Universidade Federal de Goiás, P.O. Box 131, Goiânia, Goiás 74001-970, Brazil Abstract: Virtual screening (VS) techniques are well-established tools in the modern drug discovery process, mainly used for hit finding in drug discovery. The availability of knowledge of structural information, which includes an in- creasing number of 3D protein structures and the readiness of free databases of commercially available small- molecules, provides a broad platform for VS. This review summarizes the current developments in VS regarding chemical databases and highlights the achievements as well as the challenges with an emphasis on a recent example of the successful application for the identification of new hits for sterol 14-demethylase (CYP51) of Trypanosoma cruzi. Keywords: Drug design, lead discovery, ligand-based, machine learning, pharmacophore model, performance evaluation, structure-based, virtual screening. 1. INTRODUCTION Virtual screening (VS) techniques are well-established computational tools used in the modern drug discovery process, mainly for hit finding, that is to identify new ligands for a particular molecular target [1]. Although high- throughput screening (HTS) is still a standard method for the identification of hits, it has the disadvantages of a high rate of false negatives and false positives as well as high costs, and it requires a huge investment in infrastructure [1,2]. Therefore, VS has earned prestige as a reliable alter- native to evade the costly and time-consuming process of drug discovery [3]. Unlike HTS, VS is not limited by the experimentally accessible chemical space. Therefore, VS covers a much bigger chemical space. This characteristic provides a huge advantage because finding new scaffolds appears to be more beneficial than finding many hits [4]. Although the screening capacity of computers presents difficulties re- garding the increase of chemical space coverage and the size of chemical databases, advances in computer process- ing power and computational algorithms have allowed for the implementation of VS tools and techniques to be used routinely [5,6]. The use of computational methods to screen chemical libraries rationalizes compound selection, which reduces the number of drug candidates for experimental evaluation [7]. *Address correspondence to this author at the LabMol, Faculdade de Far- mácia, Universidade Federal de Goiás, Rua 240, Qd.87, Setor Leste Univer- sitário, Goiânia – GO 74605-510, Brazil; Tel: + 55 62 3209-6451; Fax: +55 62 3209 6037; E-mail: carolina@ufg.br VS methods can be divided into ligand-based virtual screening (LBVS) and structure-based virtual screening (SBVS) [8]. Ligand-based approaches use structure-activity data from a set of known actives in order to develop models, such as similarity searching [9], machine learning [10], and ligand-based pharmacophore models [11]. On the other hand, structure-based methods use the three-dimensional (3D) structure of the biological target. In this case, candidate molecules are docked in the binding site and ranked based on their predicted binding affinity or complementarity. Another possibility is to generate pharmacophore models based on the structure of the protein [12]. The biological target can be determined experimentally by either X-ray crystallography or NMR [13], or it can be predicted computationally through homology modeling [14]. This paper focuses on current developments in virtual screening regarding chemical databases, highlights pitfalls that have strong implications in VS, and suggest possible solutions. We also present an in-house example of the suc- cessful application of VS to find new hits for sterol 14- demethylase (CYP51) of Trypanosoma cruzi. 2. VIRTUAL SCREENING WORKFLOW The VS of compound libraries is a process that can be divided into several well-defined steps. Fig. (1) presents a proposed workflow of a VS campaign using both ligand- and structure-based approaches, which will serve as the basis of the discussions in the next few sections. The VS process should begin with the design of the experiment, therefore it is necessary the knowledge of the target and the biochemical process involved. The availability of active compounds and the protein structure will indicate which