GA(M)E-QSAR: A Novel, Fully Automatic Genetic-Algorithm-(Meta)-Ensembles Approach for Binary Classification in Ligand-Based Drug Design

Abstract

Computer-aided drug design has become an important component of the drug discovery process. Despite the advances in this field, there is not a unique modeling approach that can be successfully applied to solve the whole range of problems faced during QSAR modeling. Feature selection and ensemble modeling are active areas of research in ligand-based drug design. Here we introduce the GA(M)E-QSAR algorithm that combines the search and optimization capabilities of Genetic Algorithms with the simplicity of the Adaboost ensemble-based classification algorithm to solve binary classification problems. We also explore the usefulness of Meta-Ensembles trained with Adaboost and Voting schemes to further improve the accuracy, generalization, and robustness of the optimal Adaboost Single Ensemble derived from the Genetic Algorithm optimization. We evaluated the performance of our algorithm using five data sets from the literature and found that it is capable of yielding similar or better classification results to what has been reported for these data sets with a higher enrichment of active compounds relative to the whole actives subset when only the most active chemicals are considered. More important, we compared our methodology with state of the art feature selection and classification approaches and found that it can provide highly accurate, robust, and generalizable models. In the case of the Adaboost Ensembles derived from the Genetic Algorithm search, the final models are quite simple since they consist of a weighted sum of the output of single feature classifiers. Furthermore, the Adaboost scores can be used as ranking criterion to prioritize chemicals for synthesis and biological evaluation after virtual screening experiments.

A desirability-based multi objective approach for the virtual screening discovery of broad-spectrum anti-gastric cancer agents

“…All these evidences show that ensemble modeling is a robust approach for obtaining accurate and predictive QSAR models. This observation is in agreement with similar conclusions that have been derived in previous research [ 14 , 23 , 44 , 45 ]. By following this approach it is possible to obtain predictive models for the classification-based modeling of the GC cell lines employed in this research.…”

A combined drug discovery strategy based on machine learning and molecular docking

“…As a member of Boosting family, AdaBoost is an integrated algorithm based on decision tree (Pérezcastillo et al., ). It is a type of “Ensemble Learning” where a number of weak classifiers are trained for the same training set, and then these weak classifiers are combined to form a stronger classifier.…”

“…RF can also handle both categorical and continuous variables, which can return the importance of variables and be freely implemented with high quality (Statnikov, Wang, & Aliferis, 2008). In addition, extreme random tree (ERT; Šícho, Kops, Stork, Svozil, & Kirchmair, 2017), AdaBoost (Pérezcastillo et al, 2012), gradient boosting trees (GBT) (Ericksen et al, 2017) and XGBoost (Sheridan, Wei, Liaw, Ma, & Gifford, 2016) models are also widely used in drug design and discovery and achieve favorable outcomes.…”

Systemic QSAR and phenotypic virtual screening: chasing butterflies in drug discovery

“…In-house MatLab codes were used for feature selection, modeling, and genetic algorithm optimization. Given that this stage is based on standard procedures already reported [72,82,83], the details are provided in the supplementary information online.…”