QMOD: physically meaningful QSAR

Department of Bioengineering and Therapeutic Sciences, Helen Diller Family Comprehensive Cancer Center, University of California, 1450 3rd Street, Room D373, MC 0128, P.O. Box 589001, San Francisco, CA 94158-9001, USA.
Journal of Computer-Aided Molecular Design (Impact Factor: 2.99). 10/2010; 24(10):865-78. DOI: 10.1007/s10822-010-9379-8
Source: PubMed


Computational methods for predicting ligand affinity where no protein structure is known generally take the form of regression analysis based on molecular features that have only a tangential relationship to a protein/ligand binding event. Such methods have utility in retrospective rationalization of activity patterns of substituents on a common scaffold, but are limited when either multiple scaffolds are present or when ligand alignment varies significantly based on structural changes. In addition, such methods generally assume independence and additivity of effect from scaffold substituents. Collectively, these non-physical modeling assumptions sharply limit the utility of widely used QSAR approaches for prospective prediction of ligand activity. The recently introduced Surflex-QMOD approach, by virtue of constructing physical models of binding sites, comes closer to a modeling approach that is congruent with protein ligand binding events. A set of congeneric CDK2 inhibitors showed that induced binding pockets can be quite congruent with the enzyme's active site but that model predictivity within a chemical series does not necessarily depend on congruence. Muscarinic antagonists were used to show that the QMOD approach is capable of making accurate predictions in cases where highly non-additive structure activity effects exist. The QMOD method offers a means to go beyond non-causative correlations in QSAR analysis.

Full-text preview

Available from:
  • Source
    • "Within chemical series where the effect of substituent changes is largely additive, it is difficult to discern performance differences between computationally expedient regression-based methods, moderately expensive approaches such as QMOD, or very intensive calculations such as dynamics-based simulation approaches. However, additivity occasionally breaks down quite dramatically [15, 16], and a very common case in medicinal chemistry requires predictions on molecules quite different from those upon which a model is constructed. In these cases, stark performance differences emerge among different methods. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Binding affinity prediction is frequently addressed using computational models constructed solely with molecular structure and activity data. We present a hybrid structure-guided strategy that combines molecular similarity, docking, and multiple-instance learning such that information from protein structures can be used to inform models of structure-activity relationships. The Surflex-QMOD approach has been shown to produce accurate predictions of binding affinity by constructing an interpretable physical model of a binding site with no experimental binding site structural information. We introduce a method to integrate protein structure information into the model induction process in order to construct more robust physical models. The structure-guided models accurately predict binding affinities over a broad range of compounds while producing more accurate representations of the protein pockets and ligand binding modes. Structure-guidance for the QMOD method yielded significant performance improvements, both for affinity and pose prediction, especially in cases where predictions were made on ligands very different from those used for model induction.
    Journal of Computer-Aided Molecular Design 11/2013; 27(11). DOI:10.1007/s10822-013-9688-9 · 2.99 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Computer-aided drug design is a mature field by some measures, and it has produced notable successes that underpin the study of interactions between small molecules and living systems. However, unlike a truly mature field, fallacies of logic lie at the heart of the arguments in support of major lines of research on methodology and validation thereof. Two particularly pernicious ones are cum hoc ergo propter hoc (with this, therefore because of this) and confirmation bias (seeking evidence that is confirmatory of the hypothesis at hand). These fallacies will be discussed in the context of off-target predictive modeling, QSAR, molecular similarity computations, and docking. Examples will be shown that avoid these problems.
    Journal of Computer-Aided Molecular Design 12/2011; 26(1):57-67. DOI:10.1007/s10822-011-9530-1 · 2.99 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Computational approaches for binding affinity prediction are most frequently demonstrated through cross-validation within a series of molecules or through performance shown on a blinded test set. Here, we show how such a system performs in an iterative, temporal lead optimization exercise. A series of gyrase inhibitors with known synthetic order formed the set of molecules that could be selected for "synthesis." Beginning with a small number of molecules, based only on structures and activities, a model was constructed. Compound selection was done computationally, each time making five selections based on confident predictions of high activity and five selections based on a quantitative measure of three-dimensional structural novelty. Compound selection was followed by model refinement using the new data. Iterative computational candidate selection produced rapid improvements in selected compound activity, and incorporation of explicitly novel compounds uncovered much more diverse active inhibitors than strategies lacking active novelty selection.
    Journal of Medicinal Chemistry 10/2012; 55(20):8926-42. DOI:10.1021/jm301210j · 5.45 Impact Factor
Show more