Present chapter reviews the application of Quantitative Structure-Activity Relationships for the treatment of molecules involving thousands of atoms, such as proteins, nucleic acids (DNA, RNA), or polysaccharides. This is a new developing area of interest in Chemoinformatics, and it is expected to have a growing number of applications during the forthcoming years. Among the several points to be addressed during the modeling of macromolecules, the most important one appears to be the accurate representation of the chemical structure through numerical descriptors. It has to be noticed that descriptors based on optimized three-dimensional geometry are difficult to specify, and it is also a drawback the fact that the experimental geometry is not available. However, different experts in the field have been generalizing the employment of classical types of topological descriptors in macromolecular systems.
TopIntroduction
During last decades, the Quantitative Structure-Activity Relationships (QSAR) (Hansch, 1995; Kubinyi, 2008) Theory has played an important role in many research areas, such as Medicinal Chemistry, enabling to prevent time consuming and costs associated to experiments. Since the pioneer studies performed by Hansch and Fujita in 1964 (Hansch, 1964), the QSAR formalism has been extensively applied to the study of different biological activities of interest, so the development of the theory is encouraged (Duchowicz, 2008; Duchowicz, 2009a; Duchowicz, 2009b; Goodarzi, 2009; Puzyn, 2009; Selassie, 2002).
The basis of QSAR relies on the main hypothesis that the biological activity manifested by a chemical compound completely results from its own molecular structure. It is an approach that has a thermodynamical resemblance, in the sense that QSAR is only interested on the initial and final states (molecular structure and final activity, respectively), but does not offer specific details on the usually complex mechanism/path of action involved. However, it is possible to get insight on the underlying mechanism by means of the predicted activity.
In the realms of the theory, the molecular structure is translated into the so-called molecular descriptors, describing some relevant feature of the compounds, with mathematical formulae obtained from Chemical Graph Theory, Information Theory, Quantum Mechanics, Markov Chains Theory, etc. There exist more than a thousand available descriptors in the literature (Diudea, 2001; Katritzky, 1995; Todeschini, 2009; Trinajstic, 1992), and many of these molecular descriptors are topological indices (TIs) or invariants obtained from the molecular graph, whose vertices are atoms weighted with different physicochemical properties (mass, polarity, electronegativity, charge) (Katritzky, 1993).