Improving AutoDock Vina Using Random Forest: The Growing Accuracy of Binding Affinity Prediction by the Effective Exploitation of Larger Data Sets

Abstract

There is a growing body of evidence showing that machine learning regression results in more accurate structure-based prediction of protein-ligand binding affinity. Docking methods that aim at optimizing the affinity of ligands for a target rely on how accurate their predicted ranking is. However, despite their proven advantages, machine-learning scoring functions are still not widely applied. This seems to be due to insufficient understanding of their properties and the lack of user-friendly software implementing them. Here we present a study where the accuracy of AutoDock Vina, arguably the most commonly-used docking software, is strongly improved by following a machine learning approach. We also analyse the factors that are responsible for this improvement and their generality. Most importantly, with the help of a proposed benchmark, we demonstrate that this improvement will be larger as more data becomes available for training Random Forest models, as regression models implying additive functional forms do not improve with more training data. We discuss how the latter opens the door to new opportunities in scoring function development. In order to facilitate the translation of this advance to enhance structure-based molecular design, we provide software to directly re-score Vina-generated poses and thus strongly improve their predicted binding affinity. The software is available at http://istar.cse.cuhk.edu.hk/rf-score-3.tgz and http://crcm. marseille.inserm.fr/fileadmin/rf-score-3.tgz.

Performance of machine-learning scoring functions in structure-based virtual screening

“…Many of them are provided in a way that does not permit changing the regression model, although a number of control parameters can be adjusted to tailor the SF to a particular target. Importantly, the underlying linear regression model employed by classical SFs has been shown to be unable to assimilate large amounts of structural and binding data12.…”

“…Indeed, the degree with which machine-learning SFs have outperformed classical SFs at binding affinity prediction has been highlighted by several reviews13181920. Research has been carried out on various aspects of machine-learning SFs for binding affinity prediction: how target diversity affects predictive performance21, the impact of structure-based feature selection on predictive performance22, how to build machine-learning versions of classical SFs23, how predictive performance increases with the size of the training data in both types of SFs12, how the quality of structural and binding data influences predictive performance24, which machine learning (ML) methods generate more predictive SFs25, how to correct the impact of docking pose generation error on predictive performance26 or the implementation of webservers27 and stand-alone software2628 to make these tools freely available. It is important to note that the validation of machine-learning SFs has generally been much more rigorous than that of most classical SFs13.…”

“…It is important to note that the validation of machine-learning SFs has generally been much more rigorous than that of most classical SFs13. For example, in building RF-Score v3 for binding affinity prediction, no overlapping between training and test sets is permitted by construction12. Importantly, any adjustable parameter of the machine-learning SF is selected from data not used to estimate the performance of the model13 (e.g.…”

“…We use the full DUD-E5 data sets for model building and performance assessment across 102 targets using three docking tools to generate the corresponding poses. Three machine-learning SFs using structural features with different degrees of complexity are used121422 and compared to five classical SFs. We assess the VS performance of the SFs in both established-target and novel-target settings, either tailored for broad application or for a specific target.…”

Pred-binding: large-scale protein–ligand binding affinity prediction

“…In docking method, the structures are evaluated on the basis of a force field or a scoring function 8 . It predicts the preferred conformations and binding strength of a ligand molecule, typically a small organic molecule, as bound to a protein pocket 9 . Docking provides a reasonable accuracy in predicting DTI when 3D structure of protein and large quantities of data are present 10 .…”