Descriptor Data Bank (DDB): A Cloud Platform for Multiperspective Modeling of Protein–Ligand Interactions

Abstract

Protein-ligand (PL) interactions play a key role in many life processes such as molecular recognition, molecular binding, signal transmission, and cell metabolism. Examples of interaction forces include hydrogen bonding, hydrophobic effects, steric clashes, electrostatic contacts, and van der Waals attractions. Currently, a large number of hypotheses and perspectives to model these interaction forces are scattered throughout the literature and largely forgotten. Instead, had they been assembled and utilized collectively, they would have substantially improved the accuracy of predicting binding affinity of protein-ligand complexes. In this work, we present Descriptor Data Bank (DDB), a data-driven platform on the cloud for facilitating multiperspective modeling of PL interactions. DDB is an open-access hub for depositing, hosting, executing, and sharing descriptor extraction tools and data for a large number of interaction modeling hypotheses. The platform also implements a machine-learning (ML) toolbox for automatic descriptor filtering and analysis and scoring function (SF) fitting and prediction. The descriptor filtering module is used to filter out irrelevant and/or noisy descriptors and to produce a compact subset from all available features. We seed DDB with 16 diverse descriptor extraction tools developed in-house and collected from the literature. The tools altogether generate over 2700 descriptors that characterize (i) proteins, (ii) ligands, and (iii) protein-ligand complexes. The in-house descriptors we extract are protein-specific which are based on pairwise primary and tertiary alignment of protein structures followed by clustering and trilateration. We built and used DDB's ML library to fit SFs to the in-house descriptors and those collected from the literature. We then evaluated them on several data sets that were constructed to reflect real-world drug screening scenarios. We found that multiperspective SFs that were constructed using a large number of diverse DDB descriptors capturing various PL interactions in different ways outperformed their single-perspective counterparts in all evaluation scenarios, with an average improvement of more than 15%. We also found that our proposed protein-specific descriptors improve the accuracy of SFs.

From machine learning to deep learning: Advances in scoring functions for protein–ligand docking

“…Ashtawy and Mahapatra proposed three XGBoost‐based SFs, namely BT‐Score, BT‐Dock and BT‐Screen, which were designed for binding affinity prediction, binding poses prediction and VS, respectively. These SFs were developed based on ~2,700 multiperspective descriptors generated by Descriptor Data Bank . The PDBbind v2014 was used as the major training set, and a number of computer‐generated ligand conformations and inactive protein–ligand complexes were added into the training sets for BT‐Dock and BT‐Screen, respectively.…”

“…But a recent study found that due to the similarity of the artificial decoys for different targets, the models based on even some pure ligand‐based descriptors could also bear excellent discrimination capabilities . Therefore, to solve the above problems, some recently developed platforms such as Open Drug Discovery Toolkit (ODDT) and MoleculeNet may be utilized for reference, and a comprehensive assessment towards existing ML‐based SFs with the newly developed benchmark is also required. Secondly, only few ML‐based SFs have been integrated into the popular docking tools, and most of them just tend to be used as rescoring tools, in which classical SFs are still irreplaceable so that reasonable poses can be generated first.…”

Selecting machine-learning scoring functions for structure-based virtual screening

“…lowest K d ), which would permit identifying the SFs prone to discover the most potent ligands of the target [25].     https://github.com/HongjianLi/RF-Score MIEC-SVM [54]     http://wanglab.ucsd.edu/MIEC-SVM ODDT [55]     https://github.com/oddt/oddt Descriptor Data Bank [56]     http://www.descriptordb.com…”

“…. For instance, the Descriptor Data Bank (DDB)[56] implements a ML toolbox for automatic filtering and analysis of descriptors (features) as well as SF training and prediction. The descriptor filtering module can filter out irrelevant or noisy descriptors to produce a compact subset from the 2,700 descriptors that are initially considered.…”

Machine‐learning scoring functions for structure‐based virtual screening

“…To calculate the features of each data instance, there are now several choices (Table 3). For example, Ashtawy and Mahapatra 82 presented descriptor data bank (DDB), a data‐driven platform on the cloud for facilitating multi‐perspective modeling of protein–ligand interactions. DDB enables depositing, hosting, executing, and sharing descriptor extraction tools and data for a large number of interaction modeling hypotheses.…”