The Order Theory in QSPR-QSAR Studies

DUCHOWICZ, P. R.; CASTRO, E. A.

University of Kragujevac and Faculty of Science Kragujevac

Lugar: Kragujevac; Año: 2008 p. 200

978-86-81829-90-5

Most of data analysis applications usually attempt to fit a proposed model to a set of experimental measurements or observations. There are a wide number of reasons for performing such task. For instance, the model may be purely empirical and therefore employed for performing predictions and assisting new experiments. On the other hand, it can be based on some kind of law or theory, and an evaluation of such data fitness could be later employed for gaining some insight of the underlying processes. In many situations, the ability of fitting a model to a particular set of data offer new elements for formulating new hypotheses. The type of model that can be employed in a data set depends not only on its nature, but also on the use that the model would be given. Mathematical techniques serve as valuable tools to estimate a given specific data with respect to their particular properties. Whenever the main objective is to understand the behavior of macroscopic properties exhibited by physico-chemical and biological systems, that depends upon the molecular structure and chemical reactivity, approaches based on Quantum Chemical Mechanics would undoubtly have to be adopted in order to accurately represent the involved phenomena. This translates into the necessity of considering all possible interactions present in the physical system of particles. On the other hand, it is possible to forecast in this context that a main target of next years would be the accurate description of systems constituted by thousand of interacting atoms. This stems from the fact that actual quantum mechanical calculations can be solved to a good approximation whenever they involve molecules with few atoms-free of molecular interactions. Although various methods with severe approximations can be postulated for solving the problem of interrelated particles, their fuzzy nature sometimes make them to unsatisfactorily explain the quality of the final results achieved. All what has been stated leads to the following question: is it possible to predict accurately the properties exhibited by a collection of interacting molecules? A possible answer is encompassed in the field of the Quantitative Structure Property-Activity Relationships (QSPR-QSAR) Theory. There exists a great amount of information on Ranking Methods provided in the literature, a new appealing methodology that is generally dealt in a so mathematical fashion and that need to develop more its predictive capability. Ranking Methods can be applied to a wide range of scientific fields, such as Decision Support, Toxicology, Environmental Research, Proteomics and Genomics, Analytical Chemistry, Food Chemistry, QSPR-QSAR, etc. The scope of present monograph is to explore Partial Order Ranking ideas and to provide the non-specialist reader with insights to the formalism, with special emphasis on QSPR-QSAR studies. During the last couple of years, Partial Order Theory has been used as a new tool in QSPR-QSAR studies besides the well established conventional statistical methods. Therefore it seems appropriate to review the state-of-art in this developing field. Several concepts are explained in this monograph, revising the research developed by different experts in the topic with certain degree of details, and also including illustrative application examples in order to better understand and provide insight of the methodology employed.