Application of Modelling Techniques

Molplex Ltd, i6, Charlotte Square, Newcastle Upon Tyne, UK 

RSC Drug Discovery Series No. 13 Drug Design Strategies Quantitative Approaches Edited by David J. Livingstone and Andrew M. Davis Royal Society of Chemistry 2012 

There has been a profound shift in the capabilities available to us as QSAR modellers, particularly in the expansion of readily calculated chemical descriptors and data mining methods, but at its heart remains the question of purpose. Are we seeking models that the human expert can understand and interpret Or models that are useful, validated and predictive It is easy to say both, of course, but the simplicity required for a human to understand a model can significantly reduce the chance of finding a model, and the use of powerful but complex learning methods can deliver valid and predictive models which are not interpretable in any chemical sense. 

The most important question to ask prior to starting work on a QSAR modelling exercise is what decision will be taken based on the model results What are you trying to achieve and what data do you have  

This chapter does not provide an answer to the question, often asked of experienced QSAR specialists, of what is the best modelling method to use. There isn t an answer to this question. The goal is to describe a framework for understanding the options available, and to point to more recent work that seeks to identify more rigorous quantifiable metrics for assessing performance of data mining techniques in QSAR modelling. 

---

One further question that has a substantial impact on the application of modeling techniques to biomedical problems is the choice of the design. Suppose that in our Gompertz tumor growth example we wanted to decide, given the results of some pilot experiments, when it is most useful to observe the tumor volume. In other words, we wish to choose the time points at which we obtain tumor volume observations in order to maximize the precision of the resulting parameter estimates. 

Andrea Manca is Research Fellow at the Centre for Health Economics, University of York. His research interests lie in the investigation of methodological and theoretical issues related to two broad areas the application of modelling techniques to support the decisionmaking process in health care, and the use of analytical methods in the conduction of economic analysis of health care interventions. Andrea s applied work focuses on a number of different technologies in several clinical areas, including mental health. 

In this section, we describe some topical recent examples of the applications of modeling techniques in solid-state chemistry. Our account cannot be detailed, but we aim to give an impression of the range of applicability of these techniques. 

It was originally conceived through the application of modelling techniques to the crystal structure of influenza virus NA complexed with sialic add [34,38]. 

This article describes the current state of the art and various applications of modeling techniques to the prediction of the native, biologically active conformation of a protein. This is a very active area of computational biology, and there are a number of excellent reviews of the field. " Here, we focus on the use of simplified or reduced protein models and the insights that they can provide into protein structure prediction and the nature of interactions in globular proteins. 

In this section we briefly summarize a few modern applications of simulation techniques for the understanding of crystal growth of more complex materials. In principle, liquid crystals and colloids also belong to this class, but since the relative length of their basic elements in units of their diameter is still of order about unity in contrast to polymers, for example, they can be described rather well by the more conventional models and methods as discussed above. 

The aim of the series is to present the latest fundamental material for research chemists, lecturers and students across the breadth of the subject, reaching into the various applications of theoretical techniques and modelling. The series concentrates on teaching the fundamentals of chemical structure, symmetry, bonding, reactivity, reaction mechanism, solid-state chemistry and applications in molecular modelling. It will emphasize the transfer of theoretical ideas and results to practical situations so as to demonstrate the role of theory in the solution of chemical problems in the laboratory and in industry. 

The various copolymerization models that appear in the literature (terminal, penultimate, complex dissociation, complex participation, etc.) should not be considered as alternative descriptions. They are approximations made through necessity to reduce complexity. They should, at best, be considered as a subset of some overall scheme for copolymerization. Any unified theory, if such is possible, would have to take into account all of the factors mentioned above. The models used to describe copolymerization reaction mechanisms arc normally chosen to be the simplest possible model capable of explaining a given set of experimental data. They do not necessarily provide, nor are they meant to be, a complete description of the mechanism. Much of the impetus for model development and drive for understanding of the mechanism of copolymerization conies from the need to predict composition and rates. Developments in models have followed the development and application of analytical techniques that demonstrate the inadequacy of an earlier model. 

The application of optimisation techniques for parameter estimation requires a useful statistical criterion (e.g., least-squares). A very important criterion in non-linear parameter estimation is the likelihood or probability density function. This can be combined with an error model which allows the errors to be a function of the measured value. A simple but flexible and useful error model is used in SIMUSOLV (Steiner et al., 1986 Burt, 1989). 

Capacitance and surface tension measurements have provided a wealth of data about the adsorption of ions and molecules at electrified liquid-liquid interfaces. In order to reach an understanding on the molecular level, suitable microscopic models have had to be considered. Interpretation of the capacitance measurements has been often complicated by various instrumental artifacts. Nevertheless, the results of both experimental approaches represent the reference basis for the application of other techniques of surface analysis. 

As in other fields of nanosdence, the application of STM techniques to the study of ultrathin oxide layers has opened up a new era of oxide materials research. New emergent phenomena of structure, stoichiometry, and associated physical and chemical properties have been observed and new oxide phases, hitherto unknown in the form of bulk material, have been deteded in nanolayer form and have been elucidated with the help of the STM. Some of these oxide nanolayers are and will be of paramount interest to the field of advanced catalysis, as active and passive layers in catalytic model studies, on the one hand, and perhaps even as components in real nanocatalytic applications, on the other hand. We have illustrated with the help of prototypical examples the growth and the structural variety of oxide nanolayers on metal surfaces as seen from the perspective of the STM. The selection of the particular oxide systems presented here refleds in part their relevance in catalysis and is also related to our own scientific experience. 

In references 38 and 39, Comba and Hambley introduce their topic in three parts (1) basic concepts of molecular mechanics, (2) applications of the techniques and difficulties encountered, and (3) a guide to molecular modeling of a new system. Only the introductory section of reference 38 is summarized here. 

In comparison to most other methods in surface science, STM offers two important advantages STM gives local information on the atomic scale and it can do so in situ [51]. As STM works best on flat surfaces, applications of the technique in catalysis concern models for catalysts, with the emphasis on metal single crystals. A review by Besenbacher gives an excellent overview of the possibilities [52], Nevertheless, a few investigations on real catalysts have been reported also, for example on the iron ammonia synthesis catalyst, on which... 

---