Assessing Different Classification Methods for Virtual Screening

Abstract

How well do different classification methods perform in selecting the ligands of a protein target out of large compound collections not used to train the model? Support vector machines, random forest, artificial neural networks, k-nearest-neighbor classification with genetic-algorithm-optimized feature selection, trend vectors, naïve Bayesian classification, and decision tree were used to divide databases into molecules predicted to be active and those predicted to be inactive. Training and predicted activities were treated as binary. The database was generated for the ligands of five different biological targets which have been the object of intense drug discovery efforts: HIV-reverse transcriptase, COX2, dihydrofolate reductase, estrogen receptor, and thrombin. We report significant differences in the performance of the methods independent of the biological target and compound class. Different methods can have different applications; some provide particularly high enrichment, others are strong in retrieving the maximum number of actives. We also show that these methods do surprisingly well in predicting recently published ligands of a target on the basis of initial leads and that a combination of the results of different methods in certain cases can improve results compared to the most consistent method.

The interactome: Predicting the protein-protein interactions in cells

“…Up to now, all the computational algorithms have only used single machine learning methods for the analysis and prediction of protein-protein interactions [156][157][158][159][160], or the statistical analysis of interacting patches of protein surfaces [75,149,161,162]. Our experience clearly supports the idea that each machine learning algorithm performs better for selected types of training data [163,164]. Some have very high specificity, others focus more on sensitivity.…”

Shape Signatures: New Descriptors for Predicting Cardiotoxicity In Silico

“…This technique has already gained recognition as one of the most robust and efficient classifiers (21,(56)(57)(58)63). It can tackle nontrivial problems by projecting the original descriptor vectors to a higher dimensional feature space where a clearer division between the two classes of data becomes feasible.…”

Enhancing the Effectiveness of Ligand‐Based Virtual Screening Using Data Fusion

“…We note two such recent studies. Plewczynski et al report the use of support vector machines, random forests, neural networks, k-nearest neighbour classification, trend vectors, naive Bayesian classification and decision trees to categorise sets of ligands for five pharmaceutically important biological targets [79]. Their study focused on the performance of the individual methods but they also investigated how voting might affect performance.…”