Hierarchical virtual screening for the discovery of new molecular scaffolds in antibacterial hit identification

Abstract

One of the initial steps of modern drug discovery is the identification of small organic molecules able to inhibit a target macromolecule of therapeutic interest. A small proportion of these hits are further developed into lead compounds, which in turn may ultimately lead to a marketed drug. A commonly used screening protocol used for this task is high-throughput screening (HTS). However, the performance of HTS against antibacterial targets has generally been unsatisfactory, with high costs and low rates of hit identification. Here, we present a novel computational methodology that is able to identify a high proportion of structurally diverse inhibitors by searching unusually large molecular databases in a time-, cost- and resource-efficient manner. This virtual screening methodology was tested prospectively on two versions of an antibacterial target (type II dehydroquinase from Mycobacterium tuberculosis and Streptomyces coelicolor), for which HTS has not provided satisfactory results and consequently practically all known inhibitors are derivatives of the same core scaffold. Overall, our protocols identified 100 new inhibitors, with calculated Ki ranging from 4 to 250 μM (confirmed hit rates are 60% and 62% against each version of the target). Most importantly, over 50 new active molecular scaffolds were discovered that underscore the benefits that a wide application of prospectively validated in silico screening tools is likely to bring to antibacterial hit identification.

Machine‐Learning Techniques Applied to Antibacterial Drug Discovery

“…(53). They initially identified ~4,000 compounds with shapes similar to those of known ligands from among the nine million compounds in the ZINC repository (54).…”

Neural-Network Scoring Functions Identify Structurally Novel Estrogen-Receptor Ligands

“…17 However, with some notable exceptions (see, for example, refs. 21–23 ), these kinds of functions have not been extensively used to prospectively identify novel ligands, as required for drug discovery.…”

The Importance of the Regression Model in the Structure-Based Prediction of Protein-Ligand Binding

“…For instance, RF-Score [4], the first scoring function using Random Forest (RF) [7] as the regression model, was found to outperform a range of widely-used classical scoring functions by a large margin. RF-Score has recently been used [2] to discover a large number of innovative binders of antibacterial targets. This machine-learning scoring function has now been incorporated [9] into a large-scale docking tool for prospective virtual screening, which is freely available at http://istar.cse.cuhk.edu.hk/idock/.…”

“…However, there have been a few criticisms as well. For example, the use of oversimplified features in the original version of RF-Score has been pointed out as suboptimal [14], although it is worth noting that this fact did not prevent the method from outperforming classical scoring functions [4] or achieving high hit rates in prospective virtual screening [2]. Furthermore, the combination of machine learning and these features was claimed to learn target properties, which would hamper generalisation to test set complexes with targets dissimilar from those in the training set, a conjecture that was subsequently rebutted [5].…”