Interpretation and Application of a QSAR Equation

The ultimate value of a QSAR model, of course, depends on the impact that it has on the direction of the project. The most obvious use of a QSAR is in predicting the activity of molecules not yet synthesised. Prediction can be interpolative or extrapolative [Tute 1990]. An interpolative prediction is one within the range of properties of the set of molecules used to derive the QSAR; an extrapolative prediction is one beyond it. Interpolative prediction is generally more reliable than extrapolative prediction. Inevitably, perhaps, one is most interested in predictions that improve the potency or other biological properties of the molecules and these are often extrapolative (e.g. if the correlation is linear with the physicochemical parameters). Nevertheless, there are many examples where this has been possible [Fujita 1990]. In general, one should proceed gradually until the maximum potency can be achieved. In addition to moving a project forward, a QSAR model may also help one to determine when a particular series is unlikely to lead to the desired goal. Resources can therefore be shifted to explore other areas.

There have been some efforts to quantify the accuracy of the predictions from QSAR models. This question often arises in the context of "global" and "local" models [Brown and Lewis 2006]. The data sets used for global models are often large and heterogeneous whereas those for local models tend to be smaller and more homogeneous. Many of the earlier applications of QSAR techniques were best considered as local models, being restricted to single chemical series, whereas there is now significant interest in generating global models, often of properties relating to physicochemical and ADME phenomena. The concept of model domain has proved to be useful when assessing the prediction accuracy of such global models. A simple but effective measure was found to be the similarity to the molecules in the training set, as measured by the distance to the nearest neighbour and the number of nearest neighbours [Sheridan et al. 2004]. An extension of these ideas enables one to apply a quantitative correction factor based on this "distance to model" to the prediction from a QSAR model and thereby improve its accuracy [Xu and Gao 2003; Todeschini et al. 2004; Bruneau and McElroy 2006].

A QSAR model may provide insights into the factors relevant for the activity of a series of molecules and the mechanisms underlying the activity. Indeed, this is often the most important part of a QSAR analysis. As one might expect, most benefit derives from models based on chemically meaningful and comprehensible descriptors. The non-linear dependency on logP as originally introduced by Hansch is one such example. Another example is a QSAR study on the inhibition of alcohol dehydrogenase (ADH) by 4-substituted pyrazoles (Figure 4-3) [Hansch et al. 1986; Hansch and Klein 1986]. In this case it was possible to confirm the interpretation of the QSAR equation by subsequent examination of the protein-ligand x-ray structures. The following QSAR equation was derived for the inhibition of rat liver ADH:

where ameta is the Hammett constant for meta substitutents and Ki is the enzyme inhibition constant. The negative Hammett parameter is consistent with the fact that the pyrazole group binds to the catalytic zinc atom in the enzyme; electron-releasing substituents X will increase the electron density on the nitrogen and increase binding. The near-unity coefficient of logP was interpreted as meaning that the partitioning of X between water and the enzyme parallels that between water and octanol, with complete desolvation of the substituent. The x-ray structure revealed that the substituent occupied a long, channel-like hydrophobic pocket and that there would therefore indeed be expected to be complete desolvation on binding.

Figure 4-3. Generic structure of 4-pyrazole inhibitors of alcohol dehydrogenase.

Fitness Resolution Fortress

Fitness Resolution Fortress

Learning About Fitness Resolution Fortress Can Have Amazing Benefits For Your Life And Success! Start Planning To Have Excellent Health And Fitness Today!

Get My Free Ebook


  • rhiannon
    How to inteprete qsar equations?
    2 years ago

Post a comment