Localized Heuristic Inverse Quantitative Structure Activity Relationship with Bulk Descriptors Using Numerical Gradients

Abstract

State-of-the-art quantitative structure-activity relationship (QSAR) models are often based on nonlinear machine learning algorithms, which are difficult to interpret. From a pharmaceutical perspective, QSARs are used to enhance the chemical design process. Ultimately, they should not only provide a prediction but also contribute to a mechanistic understanding and guide modifications to the chemical structure, promoting compounds with desirable biological activity profiles. Global ranking of descriptor importance and inverse QSAR have been used for these purposes. This paper introduces localized heuristic inverse QSAR, which provides an assessment of the relative ability of the descriptors to influence the biological response in an area localized around the predicted compound. The method is based on numerical gradients with parameters optimized using data sets sampled from analytical functions. The heuristic character of the method reduces the computational requirements and makes it applicable not only to fragment based methods but also to QSARs based on bulk descriptors. The application of the method is illustrated on congeneric QSAR data sets, and it is shown that the predicted influential descriptors can be used to guide structural modifications that affect the biological response in the desired direction. The method is implemented into the AZOrange Open Source QSAR package. The current implementation of localized heuristic inverse QSAR is a step toward a generally applicable method for elucidating the structure activity relationship specifically for a congeneric region of chemical space when using QSARs based on bulk properties. Consequently, this method could contribute to accelerating the chemical design process in pharmaceutical projects, as well as provide information that could enhance the mechanistic understanding for individual scaffolds.

Structural and Physico-Chemical Interpretation (SPCI) of QSAR Models and Its Comparison with Matched Molecular Pair Analysis

“…Recent trends in QSAR modeling, on the other hand, demonstrate the clear value of model interpretation to better understand structure–property relationships. , Different techniques were developed in the past to interpret specific RF, , SVM , and neural net , models. Several method-independent approaches based on calculation of partial derivatives or local gradients of descriptors − have been suggested. The main idea of these approaches is to assess the contribution of descriptors and then to transfer this knowledge to the structural level, e.g., by color coding. ,− The latter requires the use of only interpretable descriptors (e.g., fragmental descriptors, group signatures, etc.…”

Inverse‐QSPR for de novo Design: A Review

“…Stålring et al . introduced their “localized heuristic inverse QSAR”, a method able to both identify the most participating bulk descriptors in a QSPR model and to suggest the way to modify them, by advising to change (sometimes the increase or decrease precision is given) some values to reach the desired property value. According to those remarks and to their knowledge, chemists can modify the original structure in order to improve the descriptors in the recommended way.…”

Comparison of the Predictive Performance and Interpretability of Random Forest and Linear Models on Benchmark Data Sets

“…The local gradients approach is analogous to the approaches presented in a number of recent publications for interpreting nonlinear QSARs. 2,18,20,33 For brevity, we refer to the first and second approaches as the "Kuz'min/Palczewska approach" and the "local gradients approach", respectively.…”

“…Specifically, we compare two different approaches for interpreting the same Random Forest prediction. The first approach ,,− was originally proposed for Random Forest regression models and was then extended to binary classification models. , The second approach is based on estimating local gradients (i.e., vectors of partial derivatives) of the prediction with respect to the descriptors and is analogous to algorithms recently used to interpret various kinds of nonlinear QSAR models. ,,, We evaluate these approaches by means of novel scoring schemes for assessing heat map images representing the influence of molecular substructures on model predictions. We extend the heat map approach introduced by Rosenbaum et al to work with any linear or nonlinear classification or regression model, built using the same descriptors as per their earlier work, and present an Open Source implementation of our extension.…”