Property-Unmatched Decoys in Docking Benchmarks

Integrating Machine Learning-Based Pose Sampling with Established Scoring Functions for Virtual Screening

Physics-based docking methods have long been the cornerstone of structure-based virtual screening (VS). However, the emergence of machine learning (ML)-based docking approaches has opened new possibilities for enhancing VS technologies. In this study, we explore the integration of DiffDock-L, a leading ML-based pose sampling method, into VS workflows by combining it with the Vina, Gnina, and RTMScore scoring functions. We assess this integrated approach in terms of its VS effectiveness, pose sampling quality, and complementarity to traditional physics-based docking methods, such as AutoDock Vina. Our findings from the DUDE-Z benchmark dataset show that DiffDock-L performs competitively in both VS performance and pose sampling in cross-docking settings. In most cases, it generates physically plausible and biologically relevant poses, establishing itself as a viable alternative to physics-based docking algorithms. Additionally, we found that the choice of scoring function significantly influences VS success.

Improving the Hit Rates of Virtual Screening by Active Learning from Bioactivity Feedback

Virtual screening has been widely used to identify potential hit candidates that can bind to the target protein in drug discovery. Contemporary screening methods typically rely on oversimplified scoring functions, frequently yielding one-digit hit rates (or even zero) among top-ranked candidates. The substantial cost of laboratory validation further constrains the exploration of candidate molecules. We find that test-time prediction refinement is almost blank in this area, which means bioactivity feedback in the wet-lab experiments is neglected. Here, we introduce an Active Learning from Bioactivity Feedback (ALBF) framework to enhance the weak hit rate of current virtual screening methods. ALBF spends the budget of wet-lab experiments iteratively and leverages the target-specific bioactivity insights from current wet-lab tests to refine the score results (i.e., rankings). Our framework consists of two components: a novel query strategy that considers the evaluation quality and its overall influence on other top-scored molecules; and an efficient score optimization strategy that propagates the bioactivity feedback to structurally similar molecules. We evaluated ALBF on diverse subsets of the well-known DUD-E and LIT-PCBA benchmarks. Our active learning protocol averagely enhances top-100 hit rates by 60% and 30% on DUD-E and LIT-PCBA with 50 to 200 bioactivity queries on the selected molecules that are deployed in ten rounds. The consistently superior performance demonstrates ALBF's potential to enhance both the accuracy and cost-effectiveness of active learning-based laboratory testing.

A Database for Large-Scale Docking and Experimental Results

The rapid expansion of readily accessible compounds over the past six years has transformed molecular docking, improving hit rates and affinities. While many millions of molecules may score well in a docking campaign, the results are rarely fully shared, hindering the benchmarking of machine learning and chemical space exploration methods that seek to explore the expanding chemical spaces. To address this gap, we develop a website providing access to recent large library campaigns, including poses, scores, and in vitro results for campaigns against 11 targets, with 6.3 billion molecules docked and 3729 compounds experimentally tested. In a simple proof-of-concept study that speaks to the new library's utility, we use the new database to train machine learning models to predict docking scores and to find the top 0.01% scoring molecules while evaluating only 1% of the library. Even in these proof-of-concept studies, some interesting trends emerge: unsurprisingly, as models train on larger sets, they perform better; less expectedly, models could achieve high correlations with docking scores and yet still fail to enrich the new docking-discovered ligands, or even the top 0.01% of docking-ranked molecules. It will be interesting to see how these trends develop for methods more sophisticated than the simple proof-of-concept studies undertaken here; the database is openly available at lsd.docking.org.

StructureNet: Physics-Informed Hybridized Deep Learning Framework for Protein-Ligand Binding Affinity Prediction

Accurately predicting protein-ligand binding affinity is an important step in the drug discovery process. Deep learning (DL) methods have improved binding affinity prediction by using diverse categories of molecular data. However, many models rely heavily on interaction and sequence data, which impedes proper learning and limits performance in de novo applications. To address these limitations, we developed a novel graph neural network model, called StructureNet (structure-based graph neural network), to predict protein-ligand binding affinity. StructureNet improves existing DL methods by focusing entirely on structural descriptors to mitigate data memorization issues introduced by sequence and interaction data. StructureNet represents the protein and ligand structures as graphs, which are processed using a GNN-based ensemble deep learning model. StructureNet achieved a PCC of 0.68 and an AUC of 0.75 on the PDBBind v.2020 Refined Set, outperforming similar structure-based models. External validation on the DUDE-Z dataset showed that StructureNet can effectively distinguish between active and decoy ligands. Further testing on a small subset of well-known drugs indicates that StructureNet has high potential for rapid virtual screening applications. We also hybridized StructureNet with interaction- and sequence-based models to investigate their impact on testing accuracy and found minimal difference (0.01 PCC) between merged models and StructureNet as a standalone model. An ablation study found that geometric descriptors were the key drivers of model performance, with their removal leading to a PCC decrease of over 15.7%. Lastly, we tested StructureNet on ensembles of binding complex conformers generated using molecular dynamics (MD) simulations and found that incorporating multiple conformations of the same complex often improves model accuracy by capturing binding site flexibility. Overall, the results show that structural data alone are sufficient for binding affinity predictions and can address pattern recognition challenges introduced by sequence and interaction features. Additionally, structural representations of protein-ligand complexes can be considerably improved using geometric and topological descriptors. We made StructureNet GUI interface freely available online.

Virtual library docking for cannabinoid-1 receptor agonists with reduced side effects

Virtual library docking can reveal unexpected chemotypes that complement the structures of biological targets. Seeking agonists for the cannabinoid-1 receptor (CB1R), we dock 74 million tangible molecules and prioritize 46 high ranking ones for de novo synthesis and testing. Nine are active by radioligand competition, a 20% hit-rate. Structure-based optimization of one of the most potent of these (Ki = 0.7 µM) leads to '1350, a 0.95 nM ligand and a full CB1R agonist of Gi/o signaling. A cryo-EM structure of '1350 in complex with CB1R-Gi1 confirms its predicted docked pose. The lead agonist is strongly analgesic in male mice, with a 2-20-fold therapeutic window over hypolocomotion, sedation, and catalepsy and no observable conditioned place preference. These findings suggest that unique cannabinoid chemotypes may disentangle characteristic cannabinoid side-effects from analgesia, supporting the further development of cannabinoids as pain therapeutics.

A database for large-scale docking and experimental results

The rapid expansion of readily accessible compounds over the past six years has transformed molecular docking, improving hit rates and affinities. While many millions of molecules may score well in a docking campaign, the results are rarely fully shared, hindering the benchmarking of machine learning and chemical space exploration methods that seek to explore the expanding chemical spaces. To address this gap, we develop a website providing access to recent large library campaigns, including poses, scores, and in vitro results for campaigns against 11 targets, with 6.3 billion molecules docked and 3729 compounds experimentally tested. In a simple proof-of-concept study that speaks to the new library's utility, we use the new database to train machine learning models to predict docking scores and to find the top 0.01% scoring molecules while evaluating only 1% of the library. Even in these proof-of-concept studies, some interesting trends emerge: unsurprisingly, as models train on larger sets, they perform better; less expected, models could achieve high correlations with docking scores and yet still fail to enrich the new docking-discovered ligands, or even the top 0.01% of docking-ranked molecules. It will be interesting to see how these trends develop for methods more sophisticated than the simple proof-of-concept studies undertaken here; the database is openly available at lsd.docking.org.

Bayesian Optimization over Multiple Experimental Fidelities Accelerates Automated Discovery of Drug Molecules

Different experiments of differing fidelities are commonly used in the search for new drug molecules. In classic experimental funnels, libraries of molecules undergo sequential rounds of virtual, coarse, and refined experimental screenings, with each level balanced between the cost of experiments and the number of molecules screened. Bayesian optimization offers an alternative approach, using iterative experiments to locate optimal molecules with fewer experiments than large-scale screening, but without the ability to weigh the costs and benefits of different types of experiments. In this work, we combine the multifidelity approach of the experimental funnel with Bayesian optimization to search for drug molecules iteratively, taking full advantage of different types of experiments, their costs, and the quality of the data they produce. We first demonstrate the utility of the multifidelity Bayesian optimization (MF-BO) approach on a series of drug targets with data reported in ChEMBL, emphasizing what properties of the chemical search space result in substantial acceleration with MF-BO. Then we integrate the MF-BO experiment selection algorithm into an autonomous molecular discovery platform to illustrate the prospective search for new histone deacetylase inhibitors using docking scores, single-point percent inhibitions, and dose-response IC50 values as low-, medium-, and high-fidelity experiments. A chemical search space with appropriate diversity and fidelity correlation for use with MF-BO was constructed with a genetic generative algorithm. The MF-BO integrated platform then docked more than 3,500 molecules, automatically synthesized and screened more than 120 molecules for percent inhibition, and selected a handful of molecules for manual evaluation at the highest fidelity. Many of the molecules screened have never been reported in any capacity. At the end of the search, several new histone deacetylase inhibitors were found with submicromolar inhibition, free of problematic hydroxamate moieties that constrain the use of current inhibitors.

GNNSeq: A Sequence-Based Graph Neural Network for Predicting Protein-Ligand Binding Affinity

Background/Objectives: Accurately predicting protein-ligand binding affinity is essential in drug discovery for identifying effective compounds. While existing sequence-based machine learning models for binding affinity prediction have shown potential, they lack accuracy and robustness in pattern recognition, which limits their generalizability across diverse and novel binding complexes. To overcome these limitations, we developed GNNSeq, a novel hybrid machine learning model that integrates a Graph Neural Network (GNN) with Random Forest (RF) and XGBoost. Methods: GNNSeq predicts ligand binding affinity by extracting molecular characteristics and sequence patterns from protein and ligand sequences. The fully optimized GNNSeq model was trained and tested on subsets of the PDBbind dataset. The novelty of GNNSeq lies in its exclusive reliance on sequence features, a hybrid GNN framework, and an optimized kernel-based context-switching design. By relying exclusively on sequence features, GNNSeq eliminates the need for pre-docked complexes or high-quality structural data, allowing for accurate binding affinity predictions even when interaction-based or structural information is unavailable. The integration of GNN, XGBoost, and RF improves GNNSeq performance by hierarchical sequence learning, handling complex feature interactions, reducing variance, and forming a robust ensemble that improves predictions and mitigates overfitting. The GNNSeq unique kernel-based context switching scheme optimizes model efficiency and runtime, dynamically adjusts feature weighting between sequence and basic structural information, and improves predictive accuracy and model generalization. Results: In benchmarking, GNNSeq performed comparably to several existing sequence-based models and achieved a Pearson correlation coefficient (PCC) of 0.784 on the PDBbind v.2020 refined set and 0.84 on the PDBbind v.2016 core set. During external validation with the DUDE-Z v.2023.06.20 dataset, GNNSeq attained an average area under the curve (AUC) of 0.74, demonstrating its ability to distinguish active ligands from decoys across diverse ligand-receptor pairs. To further evaluate its performance, we combined GNNSeq with two additional specialized models that integrate structural and protein-ligand interaction features. When tested on a curated set of well-characterized drug-target complexes, the hybrid models achieved an average PCC of 0.89, with the top-performing model reaching a PCC of 0.97. GNNSeq was designed with a strong emphasis on computational efficiency, training on 5000+ complexes in 1 h and 32 min, with real-time affinity predictions for test complexes. Conclusions: GNNSeq provides an efficient and scalable approach for binding affinity prediction, offering improved accuracy and generalizability while enabling large-scale virtual screening and cost-effective hit identification. GNNSeq is publicly available in a server-based graphical user interface (GUI) format.

Development of Receptor Desolvation Scoring and Covalent Sampling in DOCK 6: Methods Evaluated on a RAS Test Set

Molecular docking methods are widely used in drug discovery efforts. RAS proteins are important cancer drug targets, and are useful systems for evaluating docking methods, including accounting for solvation effects and covalent small molecule binding. Water often plays a key role in small molecule binding to RAS proteins, and many inhibitors─including FDA-approved drugs─covalently bind to oncogenic RAS proteins. We assembled a RAS test set, consisting of 138 RAS protein structures and 2 structures of KRAS DNA in complex with ligands. In DOCK 6, we have implemented a receptor desolvation scoring function and a covalent docking algorithm. These new features were evaluated using the test set, with pose reproduction, cross-docking, and enrichment calculations. We tested two solvation methods for generating receptor desolvation scoring grids: GIST and 3D-RISM. Using grids from GIST or 3D-RISM, water displacements are precomputed with Gaussian-weighting, and trilinear interpolation is used to speed up this scoring calculation. To test receptor desolvation scoring, we prepared GIST and 3D-RISM grids for all KRAS systems in the test set, and we compare enrichment performance with and without receptor desolvation. Accounting for receptor desolvation using GIST improves enrichment for 51% of systems and worsens enrichment for 35% of systems, while using 3D-RISM improves enrichment for 44% of systems and worsens enrichment for 30% of systems. To more rigorously test accounting for receptor desolvation using 3D-RISM, we compare pose reproduction with and without 3D-RISM receptor desolvation. Pose reproduction docking with 3D-RISM yields a 1.8 ± 2.41% increase in success rate compared to docking without 3D-RISM. Accounting for receptor desolvation provides a small, but significant, improvement in both enrichment and pose reproduction for this set. We tested the covalent attach-and-grow algorithm on 70 KRAS systems containing covalent ligands, obtaining similar pose reproduction success rates between covalent and noncovalent docking. Comparing covalent docking to noncovalent docking, there is a 2.4 ± 3.29% increase and a 1.27 ± 3.33% decline in the success rate when docking with experimental and SMILES-generated ligand conformations, respectively. As a proof-of-concept, we performed covalent virtual screens with and without receptor desolvation scoring, targeting the switch II pocket of KRAS, using 3.4 million make-on-demand acrylamide compounds from the Enamine REAL database. On average, the attach-and-grow algorithm spends approximately 17.61 s per molecule across the screen. The test set is available at https://github.com/tbalius/teb_docking_test_sets.

Docking 14 million virtual isoquinuclidines against the mu and kappa opioid receptors reveals dual antagonists-inverse agonists with reduced withdrawal effects

Large library docking of tangible molecules has revealed potent ligands across many targets. While make-on-demand libraries now exceed 75 billion enumerated molecules, their synthetic routes are dominated by a few reaction types, reducing diversity and inevitably leaving many interesting bioactive-like chemotypes unexplored. Here, we investigate the large-scale enumeration and targeted docking of isoquinuclidines. These "natural-product-like" molecules are rare in the current libraries and are functionally congested, making them interesting as receptor probes. Using a modular, four-component reaction scheme, we built and docked a virtual library of over 14.6 million isoquinuclidines against both the μ- and κ-opioid receptors (MOR and KOR, respectively). Synthesis and experimental testing of 18 prioritized compounds found nine ligands with low μM affinities. Structure-based optimization revealed low- and sub-nM antagonists and inverse agonists targeting both receptors. Cryo-electron microscopy (cryoEM) structures illuminate the origins of activity on each target. In mouse behavioral studies, a potent member of the series with joint MOR-antagonist and KOR-inverse-agonist activity reversed morphine-induced analgesia, phenocopying the MOR-selective anti-overdose agent naloxone. Encouragingly, the new molecule induced less severe opioid-induced withdrawal symptoms compared to naloxone during withdrawal precipitation, and did not induce conditioned-place aversion, likely reflecting a reduction of dysphoria due to the compound's KOR-inverse agonism. The strengths and weaknesses of bespoke library docking, and of docking for opioid receptor polypharmacology, will be considered.

MolBiC: the cell-based landscape illustrating molecular bioactivities

The measurement of cell-based molecular bioactivity (CMB) is critical for almost every step of drug development. With the booming application of AI in biomedicine, it is essential to have the CMB data to promote the learning of cell-based patterns for guiding modern drug discovery, but no database providing such information has been constructed yet. In this study, we introduce MolBiC, a knowledge base designed to describe valuable data on molecular bioactivity measured within a cellular context. MolBiC features 550 093 experimentally validated CMBs, encompassing 321 086 molecules and 2666 targets across 988 cell lines. Our MolBiC database is unique in describing the valuable data of CMB, which meets the critical demands for CMB-based big data promoting the learning of cell-based molecular/pharmaceutical pattern in drug discovery and development. MolBiC is now freely accessible without any login requirement at: https://idrblab.org/MolBiC/.

Assessing small molecule conformational sampling methods in molecular docking

Small molecule conformational sampling plays a pivotal role in molecular docking. Recent advancements have led to the emergence of various conformational sampling methods, each employing distinct algorithms. This study investigates the impact of different small molecule conformational sampling methods in molecular docking using UCSF DOCK 3.7. Specifically, six traditional sampling methods (Omega, BCL::Conf, CCDC Conformer Generator, ConfGenX, Conformator, RDKit ETKDGv3) and a deep learning-based model (Torsional Diffusion) for generating conformational ensembles are evaluated. These ensembles are subsequently docked against the Platinum Diverse Dataset, the PoseBusters dataset and the DUDE-Z dataset to assess binding pose reproducibility and screening power. Notably, different sampling methods exhibit varying performance due to their unique preferences, such as dihedral angle sampling ranges on rotatable bonds. Combining complementary methods may lead to further improvements in docking performance.

Validation guidelines for drug-target prediction methods

INTRODUCTION

Mapping the interactions between pharmaceutical compounds and their molecular targets is a fundamental aspect of drug discovery and repurposing. Drug-target interactions are important for elucidating mechanisms of action and optimizing drug efficacy and safety profiles. Several computational methods have been developed to systematically predict drug-target interactions. However, computational and experimental validation of the drug-target predictions greatly vary across the studies.

AREAS COVERED

Through a PubMed query, a corpus comprising 3,286 articles on drug-target interaction prediction published within the past decade was covered. Natural language processing was used for automated abstract classification to study the evolution of computational methods, validation strategies and performance assessment metrics in the 3,286 articles. Additionally, a manual analysis of 259 studies that performed experimental validation of computational predictions revealed prevalent experimental protocols.

EXPERT OPINION

Starting from 2014, there has been a noticeable increase in articles focusing on drug-target interaction prediction. Docking and regression stands out as the most commonly used techniques among computational methods, and cross-validation is frequently employed as the computational validation strategy. Testing the predictions using multiple, orthogonal validation strategies is recommended and should be reported for the specific target prediction applications. Experimental validation remains relatively rare and should be performed more routinely to evaluate biological relevance of predictions.

ROSHAMBO: Open-Source Molecular Alignment and 3D Similarity Scoring

Efficient virtual screening techniques are critical in drug discovery for identifying potential drug candidates. We present an open-source package for molecular alignment and 3D similarity calculations optimized for large-scale virtual screening of small molecules. This work parallels widely used proprietary tools and offers an approach complementary to structure-based virtual screening. Our package employs the PAPER software for optimizing molecular alignments based on Gaussian volume overlaps. GPU acceleration is utilized to significantly reduce computational time and resource requirements. After obtaining the optimal alignments between the target and the query molecules, both shape and color (based on pharmacophore features) scores are computed to assess molecular similarity, with aligned molecules optionally being output in sdf format. The package was benchmarked using the DUDE-Z public data sets. Results demonstrated the package's near-state-of-the-art performance and robustness across multiple target classes, with speed that enables many routine ligand-based drug discovery workflows. As an open-source and freely available resource (github.com/molecularinformatics/roshambo) with both a convenient Python API and command line interface, our package also addresses the need for accessible and efficient virtual screening tools in drug discovery.

Introducing SpaceGA: A Search Tool to Accelerate Large Virtual Screenings of Combinatorial Libraries

The growth of make-on-demand libraries in recent years has provided completely new possibilities for virtual screening for discovering new hit compounds with specific and favorable properties. However, since these libraries now contain billions of compounds, screening them using traditional methods such as molecular docking has become challenging and requires substantial computational resources. Thus, to take real advantage of the new possibilities introduced by the make-on-demand libraries, different methods have been proposed to accelerate the screening process and prioritize molecules for evaluation. Here, we introduce SpaceGA, a genetic algorithm that leverages the rapid similarity search tool SpaceLight (Bellmann, L.; Penner, P.; Rarey, M. Topological similarity search in large combinatorial fragment spaces. J. Chem. Inf. Model. 2021, 61, 238-251). to constrain the optimization process to accessible compounds within desired combinatorial libraries. As shown herein, SpaceGA is able to efficiently identify molecules with desired properties from trillions of synthesizable compounds by enumerating and evaluating only a small fraction of them. On this basis, SpaceGA represents a promising new tool for accelerating and simplifying virtual screens of ultralarge combinatorial databases.

Data-centric challenges with the application and adoption of artificial intelligence for drug discovery

INTRODUCTION

Artificial intelligence (AI) is exhibiting tremendous potential to reduce the massive costs and long timescales of drug discovery. There are however important challenges currently limiting the impact and scope of AI models.

AREAS COVERED

In this perspective, the authors discuss a range of data issues (bias, inconsistency, skewness, irrelevance, small size, high dimensionality), how they challenge AI models, and which issue-specific mitigations have been effective. Next, they point out the challenges faced by uncertainty quantification techniques aimed at enhancing and trusting the predictions from these AI models. They also discuss how conceptual errors, unrealistic benchmarks and performance misestimation can confound the evaluation of models and thus their development. Lastly, the authors explain how human bias, whether from AI experts or drug discovery experts, constitutes another challenge that can be alleviated by gaining more prospective experience.

EXPERT OPINION

AI models are often developed to excel on retrospective benchmarks unlikely to anticipate their prospective performance. As a result, only a few of these models are ever reported to have prospective value (e.g. by discovering potent and innovative drug leads for a therapeutic target). The authors have discussed what can go wrong in practice with AI for drug discovery. The authors hope that this will help inform the decisions of editors, funders investors, and researchers working in this area.

Retrieval Augmented Docking Using Hierarchical Navigable Small Worlds

Make-on-demand chemical libraries have drastically increased the reach of molecular docking, with the enumerated ready-to-dock ZINC-22 library approaching 6.4 billion molecules (July 2024). While ever-growing libraries result in better-scoring molecules, the computational resources required to dock all of ZINC-22 make this endeavor infeasible for most. Here, we organize and traverse chemical space with hierarchical navigable small-world graphs, a method we term retrieval augmented docking (RAD). RAD recovers most virtual actives, despite docking only a fraction of the library. Furthermore, RAD is protein-agnostic, supporting additional docking campaigns without additional computational overhead. In depth, we assess RAD on published large-scale docking campaigns against D4 and AmpC spanning 99.5 million and 138 million molecules, respectively. RAD recovers 95% of DOCK virtual actives for both targets after evaluating only 10% of the libraries. In breadth, RAD shows widespread applicability against 43 DUDE-Z proteins, evaluating 50.3 million associations. On average, RAD recovers 87% of virtual actives while docking 10% of the library without sacrificing chemical diversity.

Large library docking identifies positive allosteric modulators of the calcium-sensing receptor

Positive allosteric modulator (PAM) drugs enhance the activation of the calcium-sensing receptor (CaSR) and suppress parathyroid hormone (PTH) secretion. Unfortunately, these hyperparathyroidism-treating drugs can induce hypocalcemia and arrhythmias. Seeking improved modulators, we docked libraries of 2.7 million and 1.2 billion molecules against the CaSR structure. The billion-molecule docking found PAMs with a 2.7-fold higher hit rate than the million-molecule library, with hits up to 37-fold more potent. Structure-based optimization led to nanomolar leads. In ex vivo organ assays, one of these PAMs was 100-fold more potent than the standard of care, cinacalcet, and reduced serum PTH levels in mice without the hypocalcemia typical of CaSR drugs. As determined from cryo-electron microscopy structures, the PAMs identified here promote CaSR conformations that more closely resemble the activated state than those induced by the established drugs.

Toward Unveiling Putative Binding Sites of Interleukin-33: Insights from Mixed-Solvent Molecular Dynamics Simulations of the Interleukin-1 Family

The interleukin (IL)-1 family is a major proinflammatory cytokine family, ranging from the well-studied IL-1s to the most recently discovered IL-33. As a new focus, IL-33 has attracted extensive research for its crucial immunoregulatory roles, leading to the development of notable monoclonal antibodies as clinical candidates. Efforts to develop small molecules disrupting IL-33/ST2 interaction remain highly desired but encounter challenges due to the shallow and featureless interfaces. The information from relative cytokines has shown that traditional binding site identification methods still struggle in mapping cryptic sites, necessitating dynamic approaches to uncover druggable pockets on IL-33. Here, we employed mixed-solvent molecular dynamics (MixMD) simulations with diverse-property probes to map the hotspots of IL-33 and identify potential binding sites. The protocol was first validated using the known binding sites of two IL-1 family members and then applied to the structure of IL-33. Our simulations revealed several binding sites and proposed side-chain rearrangements essential for the binding of a known inhibitor, aligning well with experimental NMR findings. Further microsecond-time scale simulations of this IL-33-protein complex unveiled distinct binding modes with varying occurrences. These results could facilitate future efforts in developing ligands to target challenging flexible pockets of IL-33 and IL-1 family cytokines in general.

Can Current Molecular Docking Methods Accurately Predict RNA Inhibitors?

Ribonucleic acids (RNAs), particularly the noncoding RNAs, play key roles in cancer, making them attractive drug targets. While conventional methods such as high throughput screening are resource-intensive, computational methods such as RNA-ligand docking can be used as an alternative. However, currently available docking methods are fine-tuned to perform protein-ligand and protein-protein docking. In this work, we evaluated three commonly used docking methods─AutoDock Vina, HADDOCK, and HDOCK─alongside RLDOCK, which is specifically designed for RNA-ligand docking. Our evaluation was based on several criteria including cognate docking, blind docking, scoring potential, and ranking potential. In cognate docking, only RLDOCK showed a success rate of 70% for the top-scoring docked pose. Despite this, all four docking methods did not achieve an overall success rate exceeding 50% amidst our attempt to refine the top-scoring docked poses using molecular dynamics simulations. Meanwhile, all four docking methods showed poor performance in scoring potential evaluation. Although AutoDock Vina achieved an area under the receiver operating characteristic curve of 0.70, it showed poor performance in terms of Matthews' correlation coefficient, precision, enrichment factors, and normalized enrichment factors at 1, 2, and 5%. These results highlight the growing need for further optimization of docking methods to assess RNA-ligand interactions.

Building shape-focused pharmacophore models for effective docking screening

The performance of molecular docking can be improved by comparing the shape similarity of the flexibly sampled poses against the target proteins' inverted binding cavities. The effectiveness of these pseudo-ligands or negative image-based models in docking rescoring is boosted further by performing enrichment-driven optimization. Here, we introduce a novel shape-focused pharmacophore modeling algorithm O-LAP that generates a new class of cavity-filling models by clumping together overlapping atomic content via pairwise distance graph clustering. Top-ranked poses of flexibly docked active ligands were used as the modeling input and multiple alternative clustering settings were benchmark-tested thoroughly with five demanding drug targets using random training/test divisions. In docking rescoring, the O-LAP modeling typically improved massively on the default docking enrichment; furthermore, the results indicate that the clustered models work well in rigid docking. The C+ +/Qt5-based algorithm O-LAP is released under the GNU General Public License v3.0 via GitHub ( https://github.com/jvlehtonen/overlap-toolkit ). SCIENTIFIC CONTRIBUTION: This study introduces O-LAP, a C++/Qt5-based graph clustering software for generating new type of shape-focused pharmacophore models. In the O-LAP modeling, the target protein cavity is filled with flexibly docked active ligands, the overlapping ligand atoms are clustered, and the shape/electrostatic potential of the resulting model is compared against the flexibly sampled molecular docking poses. The O-LAP modeling is shown to ensure high enrichment in both docking rescoring and rigid docking based on comprehensive benchmark-testing.

AlphaFold accelerated discovery of psychotropic agonists targeting the trace amine-associated receptor 1

Artificial intelligence is revolutionizing protein structure prediction, providing unprecedented opportunities for drug design. To assess the potential impact on ligand discovery, we compared virtual screens using protein structures generated by the AlphaFold machine learning method and traditional homology modeling. More than 16 million compounds were docked to models of the trace amine-associated receptor 1 (TAAR1), a G protein-coupled receptor of unknown structure and target for treating neuropsychiatric disorders. Sets of 30 and 32 highly ranked compounds from the AlphaFold and homology model screens, respectively, were experimentally evaluated. Of these, 25 were TAAR1 agonists with potencies ranging from 12 to 0.03 μM. The AlphaFold screen yielded a more than twofold higher hit rate (60%) than the homology model and discovered the most potent agonists. A TAAR1 agonist with a promising selectivity profile and drug-like properties showed physiological and antipsychotic-like effects in wild-type but not in TAAR1 knockout mice. These results demonstrate that AlphaFold structures can accelerate drug discovery.

Structure-based virtual screening of vast chemical space as a starting point for drug discovery

Structure-based virtual screening aims to find molecules forming favorable interactions with a biological macromolecule using computational models of complexes. The recent surge of commercially available chemical space provides the opportunity to search for ligands of therapeutic targets among billions of compounds. This review offers a compact overview of structure-based virtual screens of vast chemical spaces, highlighting successful applications in early drug discovery for therapeutically important targets such as G protein-coupled receptors and viral enzymes. Emphasis is placed on strategies to explore ultra-large chemical libraries and synergies with emerging machine learning techniques. The current opportunities and future challenges of virtual screening are discussed, indicating that this approach will play an important role in the next-generation drug discovery pipeline.

Artificial intelligence and cheminformatics tools: a contribution to the drug development and chemical science

In the ever-evolving field of drug discovery, the integration of Artificial Intelligence (AI) and Machine Learning (ML) with cheminformatics has proven to be a powerful combination. Cheminformatics, which combines the principles of computer science and chemistry, is used to extract chemical information and search compound databases, while the application of AI and ML allows for the identification of potential hit compounds, optimization of synthesis routes, and prediction of drug efficacy and toxicity. This collaborative approach has led to the discovery, preclinical evaluations and approval of over 70 drugs in recent years. To aid researchers in the pursuit of new drugs, this article presents a comprehensive list of databases, datasets, predictive and generative models, scoring functions and web platforms that have been launched between 2021 and 2022. These resources provide a wealth of information and tools for computer-assisted drug development, and are a valuable asset for those working in the field of cheminformatics. Overall, the integration of AI, ML and cheminformatics has greatly advanced the drug discovery process and continues to hold great potential for the future. As new resources and technologies become available, we can expect to see even more groundbreaking discoveries and advancements in these fields.Communicated by Ramaswamy H. Sarma.

Do Molecular Fingerprints Identify Diverse Active Drugs in Large-Scale Virtual Screening? (No)

Computational approaches for small-molecule drug discovery now regularly scale to the consideration of libraries containing billions of candidate small molecules. One promising approach to increased the speed of evaluating billion-molecule libraries is to develop succinct representations of each molecule that enable the rapid identification of molecules with similar properties. Molecular fingerprints are thought to provide a mechanism for producing such representations. Here, we explore the utility of commonly used fingerprints in the context of predicting similar molecular activity. We show that fingerprint similarity provides little discriminative power between active and inactive molecules for a target protein based on a known active-while they may sometimes provide some enrichment for active molecules in a drug screen, a screened data set will still be dominated by inactive molecules. We also demonstrate that high-similarity actives appear to share a scaffold with the query active, meaning that they could more easily be identified by structural enumeration. Furthermore, even when limited to only active molecules, fingerprint similarity values do not correlate with compound potency. In sum, these results highlight the need for a new wave of molecular representations that will improve the capacity to detect biologically active molecules based on their similarity to other such molecules.

The impact of Library Size and Scale of Testing on Virtual Screening

Virtual libraries for ligand discovery have recently increased 10,000-fold, and this is thought to have improved hit rates and potencies from library docking. This idea has not, however, been experimentally tested in direct comparisons of larger-vs-smaller libraries. Meanwhile, though libraries have exploded, the scale of experimental testing has little changed, with often only dozens of high-ranked molecules investigated, making interpretation of hit rates and affinities uncertain. Accordingly, we docked a 1.7 billion molecule virtual library against the model enzyme AmpC β-lactamase, testing 1,521 new molecules and comparing the results to the same screen with a library of 99 million molecules, where only 44 molecules were tested. Encouragingly, the larger screen outperformed the smaller one: hit rates improved by two-fold, more new scaffolds were discovered, and potency improved. Overall, 50-fold more inhibitors were found, supporting the idea that there are many more compounds to be discovered than are being tested. With so many compounds evaluated, we could ask how the results vary with number tested, sampling smaller sets at random from the 1521. Hit rates and affinities were highly variable when we only sampled dozens of molecules, and it was only when we included several hundred molecules that results converged. As docking scores improved, so too did the likelihood of a molecule binding; hit rates improved steadily with docking score, as did affinities. This also appeared true on reanalysis of large-scale results against the σ2 and dopamine D4 receptors. It may be that as the scale of both the virtual libraries and their testing grows, not only are better ligands found but so too does our ability to rank them.

Small vs. Large Library Docking for Positive Allosteric Modulators of the Calcium Sensing Receptor

Drugs acting as positive allosteric modulators (PAMs) to enhance the activation of the calcium sensing receptor (CaSR) and to suppress parathyroid hormone (PTH) secretion can treat hyperparathyroidism but suffer from side effects including hypocalcemia and arrhythmias. Seeking new CaSR modulators, we docked libraries of 2.7 million and 1.2 billion molecules against transforming pockets in the active-state receptor dimer structure. Consistent with simulations suggesting that docking improves with library size, billion-molecule docking found new PAMs with a hit rate that was 2.7-fold higher than the million-molecule library and with hits up to 37-fold more potent. Structure-based optimization of ligands from both campaigns led to nanomolar leads, one of which was advanced to animal testing. This PAM displays 100-fold the potency of the standard of care, cinacalcet, in ex vivo organ assays, and reduces serum PTH levels in mice by up to 80% without the hypocalcemia typical of CaSR drugs. Cryo-EM structures with the new PAMs show that they induce residue rearrangements in the binding pockets and promote CaSR dimer conformations that are closer to the G-protein coupled state compared to established drugs. These findings highlight the promise of large library docking for therapeutic leads, especially when combined with experimental structure determination and mechanism.

Large library docking for cannabinoid-1 receptor agonists with reduced side effects

Large library docking can reveal unexpected chemotypes that complement the structures of biological targets. Seeking new agonists for the cannabinoid-1 receptor (CB1R), we docked 74 million tangible molecules, prioritizing 46 high ranking ones for de novo synthesis and testing. Nine were active by radioligand competition, a 20% hit-rate. Structure-based optimization of one of the most potent of these (Ki = 0.7 uM) led to '4042, a 1.9 nM ligand and a full CB1R agonist. A cryo-EM structure of the purified enantiomer of '4042 ('1350) in complex with CB1R-Gi1 confirmed its docked pose. The new agonist was strongly analgesic, with generally a 5-10-fold therapeutic window over sedation and catalepsy and no observable conditioned place preference. These findings suggest that new cannabinoid chemotypes may disentangle characteristic cannabinoid side-effects from their analgesia, supporting the further development of cannabinoids as pain therapeutics.

DockOpt: A Tool for Automatic Optimization of Docking Models

Molecular docking is a widely used technique for leveraging protein structure for ligand discovery, but it remains difficult to utilize due to limitations that have not been adequately addressed. Despite some progress toward automation, docking still requires expert guidance, hindering its adoption by a broader range of investigators. To make docking more accessible, we developed a new utility called DockOpt, which automates the creation, evaluation, and optimization of docking models prior to their deployment in large-scale prospective screens. DockOpt outperforms our previous automated pipeline across all 43 targets in the DUDE-Z benchmark data set, and the generated models for 84% of targets demonstrate sufficient enrichment to warrant their use in prospective screens, with normalized LogAUC values of at least 15%. DockOpt is available as part of the Python package Pydock3 included in the UCSF DOCK 3.8 distribution, which is available for free to academic researchers at https://dock.compbio.ucsf.edu and free for everyone upon registration at https://tldr.docking.org.

Docking for Molecules That Bind in a Symmetric Stack with SymDOCK

Discovering ligands for amyloid fibrils, such as those formed by the tau protein, is an area of great current interest. In recent structures, ligands bind in stacks in the tau fibrils to reflect the rotational and translational symmetry of the fibril itself; in these structures, the ligands make few interactions with the protein but interact extensively with each other. To exploit this symmetry and stacking, we developed SymDOCK, a method to dock molecules that follow the protein's symmetry. For each prospective ligand pose, we apply the symmetry operation of the fibril to generate a self-interacting and fibril-interacting stack, checking that doing so will not cause a clash between the original molecule and its image. Absent a clash, we retain that pose and add the ligand-ligand van der Waals energy to the ligand's docking score (here using DOCK3.8). We can check these geometries and energies using an implementation of ANI, a neural-network-based quantum-mechanical evaluation of the ligand stacking energies. In retrospective calculations, symmetry docking can reproduce the poses of three tau PET tracers whose structures have been determined. More convincingly, in a prospective study, SymDOCK predicted the structure of the PET tracer MK-6240 bound in a symmetrical stack to AD PHF tau before that structure was determined; the docked pose was used to determine how MK-6240 fit the cryo-EM density. In proof-of-concept studies, SymDOCK enriched known ligands over property-matched decoys in retrospective screens without sacrificing docking speed and can address large library screens that seek new symmetrical stackers. Future applications of this approach will be considered.

Redocking the PDB

Molecular docking is a standard technique in structure-based drug design (SBDD). It aims to predict the 3D structure of a small molecule in the binding site of a receptor (often a protein). Despite being a common technique, it often necessitates multiple tools and involves manual steps. Here, we present the JAMDA preprocessing and docking workflow that is easy to use and allows fully automated docking. We evaluate the JAMDA docking workflow on binding sites extracted from the complete PDB and derive key factors determining JAMDA's docking performance. With that, we try to remove most of the bias due to manual intervention and provide a realistic estimate of the redocking performance of our JAMDA preprocessing and docking workflow for any PDB structure. On this large PDBScan22 data set, our JAMDA workflow finds a pose with an RMSD of at most 2 Å to the crystal ligand on the top rank for 30.1% of the structures. When applying objective structure quality filters to the PDBScan22 data set, the success rate increases to 61.8%. Given the prepared structures from the JAMDA preprocessing pipeline, both JAMDA and the widely used AutoDock Vina perform comparably on this filtered data set (the PDBScan22-HQ data set).

DOCK 6: Incorporating hierarchical traversal through precomputed ligand conformations to enable large-scale docking

To allow DOCK 6 access to unprecedented chemical space for screening billions of small molecules, we have implemented features from DOCK 3.7 into DOCK 6, including a search routine that traverses precomputed ligand conformations stored in a hierarchical database. We tested them on the DUDE-Z and SB2012 test sets. The hierarchical database search routine is 16 times faster than anchor-and-grow. However, the ability of hierarchical database search to reproduce the experimental pose is 16% worse than that of anchor-and-grow. The enrichment performance is on average similar, but DOCK 3.7 has better enrichment than DOCK 6, and DOCK 6 is on average 1.7 times slower. However, with post-docking torsion minimization, DOCK 6 surpasses DOCK 3.7. A large-scale virtual screen is performed with DOCK 6 on 23 million fragment molecules. We use current features in DOCK 6 to complement hierarchical database calculations, including torsion minimization, which is not available in DOCK 3.7.

Discovery of potential WEE1 inhibitors via hybrid virtual screening

G2/M cell cycle checkpoint protein WEE1 kinase is a promising target for inhibiting tumor growth. Although various WEE1 inhibitors have entered clinical investigations, their therapeutic efficacy and safety profile remain unsatisfactory. In this study, we employed a comprehensive virtual screening workflow, which included Schrödinger-Glide molecular docking at different precision levels, as well as the utilization of tools such as MM/GBSA and Deepdock to predict the binding affinity between targets and ligands, in order to identify potential WEE1 inhibitors. Out of ten molecules screened, 50% of these molecules exhibited strong inhibitory activity against WEE1. Among them, compounds 4 and 5 showed excellent inhibitory activity with IC50 values of 1.069 and 3.77 nM respectively, which was comparable to AZD1775. Further investigations revealed that compound 4 displayed significant anti-proliferative effects in A549, PC9, and HuH-7 cells and could also induce apoptosis and G1 phase arrest in PC9 cells. Additionally, molecular dynamics simulations unveiled the binding details of compound 4 with WEE1, notably the crucial hydrogen bond interactions formed with Cys379. In summary, this comprehensive virtual screening workflow, combined with in vitro testing and computational modeling, holds significant importance in the development of promising WEE1 inhibitors.

Docking for EP4R antagonists active against inflammatory pain

The lipid prostaglandin E2 (PGE2) mediates inflammatory pain by activating G protein-coupled receptors, including the prostaglandin E2 receptor 4 (EP4R). Nonsteroidal anti-inflammatory drugs (NSAIDs) reduce nociception by inhibiting prostaglandin synthesis, however, the disruption of upstream prostanoid biosynthesis can lead to pleiotropic effects including gastrointestinal bleeding and cardiac complications. In contrast, by acting downstream, EP4R antagonists may act specifically as anti-inflammatory agents and, to date, no selective EP4R antagonists have been approved for human use. In this work, seeking to diversify EP4R antagonist scaffolds, we computationally dock over 400 million compounds against an EP4R crystal structure and experimentally validate 71 highly ranked, de novo synthesized molecules. Further, we show how structure-based optimization of initial docking hits identifies a potent and selective antagonist with 16 nanomolar potency. Finally, we demonstrate favorable pharmacokinetics for the discovered compound as well as anti-allodynic and anti-inflammatory activity in several preclinical pain models in mice.

A practical guide to machine-learning scoring for structure-based virtual screening

Structure-based virtual screening (SBVS) via docking has been used to discover active molecules for a range of therapeutic targets. Chemical and protein data sets that contain integrated bioactivity information have increased both in number and in size. Artificial intelligence and, more concretely, its machine-learning (ML) branch, including deep learning, have effectively exploited these data sets to build scoring functions (SFs) for SBVS against targets with an atomic-resolution 3D model (e.g., generated by X-ray crystallography or predicted by AlphaFold2). Often outperforming their generic and non-ML counterparts, target-specific ML-based SFs represent the state of the art for SBVS. Here, we present a comprehensive and user-friendly protocol to build and rigorously evaluate these new SFs for SBVS. This protocol is organized into four sections: (i) using a public benchmark of a given target to evaluate an existing generic SF; (ii) preparing experimental data for a target from public repositories; (iii) partitioning data into a training set and a test set for subsequent target-specific ML modeling; and (iv) generating and evaluating target-specific ML SFs by using the prepared training-test partitions. All necessary code and input/output data related to three example targets (acetylcholinesterase, HMG-CoA reductase, and peroxisome proliferator-activated receptor-α) are available at https://github.com/vktrannguyen/MLSF-protocol , can be run by using a single computer within 1 week and make use of easily accessible software/programs (e.g., Smina, CNN-Score, RF-Score-VS and DeepCoy) and web resources. Our aim is to provide practical guidance on how to augment training data to enhance SBVS performance, how to identify the most suitable supervised learning algorithm for a data set, and how to build an SF with the highest likelihood of discovering target-active molecules within a given compound library.

Docking for molecules that bind in a symmetric stack with SymDOCK

Discovering ligands for amyloid fibrils, such as those formed by the tau protein, is an area of much current interest. In recent structures, ligands bind in stacks in the tau fibrils to reflect the rotational and translational symmetry of the fibril itself; in these structures the ligands make few interactions with the protein but interact extensively with each other. To exploit this symmetry and stacking, we developed SymDOCK, a method to dock molecules that follow the protein's symmetry. For each prospective ligand pose, we apply the symmetry operation of the fibril to generate a self-interacting and fibril-interacting stack, checking that doing so will not cause a clash between the original molecule and its image. Absent a clash, we retain that pose and add the ligand-ligand van der Waals energy to the ligand's docking score (here using DOCK3.8). We can check these geometries and energies using an implementation of ANI, a neural network-based quantum-mechanical evaluation of the ligand stacking energies. In retrospective calculations, symmetry docking can reproduce the poses of three tau PET tracers whose structures have been determined. More convincingly, in a prospective study SymDOCK predicted the structure of the PET tracer MK-6240 bound in a symmetrical stack to AD PHF tau before that structure was determined; the docked pose was used to determine how MK-6240 fit the cryo-EM density. In proof-of-concept studies, SymDOCK enriched known ligands over property-matched decoys in retrospective screens without sacrificing docking speed, and can address large library screens that seek new symmetrical stackers. Future applications of this approach will be considered.

Discovery of AD258 as a Sigma Receptor Ligand with Potent Antiallodynic Activity

The design and synthesis of a series of 2,7-diazaspiro[4.4]nonane derivatives as potent sigma receptor (SR) ligands, associated with analgesic activity, are the focus of this work. In this study, affinities at S1R and S2R were measured, and molecular modeling studies were performed to investigate the binding pose characteristics. The most promising compounds were subjected to in vitro toxicity testing and subsequently screened for in vivo analgesic properties. Compound 9d (AD258) exhibited negligible in vitro cellular toxicity and a high binding affinity to both SRs (KiS1R = 3.5 nM, KiS2R = 2.6 nM), but not for other pain-related targets, and exerted high potency in a model of capsaicin-induced allodynia, reaching the maximum antiallodynic effect at very low doses (0.6-1.25 mg/kg). Functional activity experiments showed that S1R antagonism is needed for the effects of 9d and that it did not induce motor impairment. In addition, 9d exhibited a favorable pharmacokinetic profile.

Low Molecular Weight Inhibitors Targeting the RNA-Binding Protein HuR

The RNA-binding protein human antigen R (HuR) regulates stability, translation, and nucleus-to-cytoplasm shuttling of its target mRNAs. This protein has been progressively recognized as a relevant therapeutic target for several pathologies, like cancer, neurodegeneration, as well as inflammation. Inhibitors of mRNA binding to HuR might thus be beneficial against a variety of diseases. Here, we present the rational identification of structurally novel HuR inhibitors. In particular, by combining chemoinformatic approaches, high-throughput virtual screening, and RNA-protein pulldown assays, we demonstrate that the 4-(2-(2,4,6-trioxotetrahydropyrimidin-5(2H)-ylidene)hydrazineyl)benzoate ligand exhibits a dose-dependent HuR inhibition effect in binding experiments. Importantly, the chemical scaffold is new with respect to the currently known HuR inhibitors, opening up a new avenue for the design of pharmaceutical agents targeting this important protein.

Large library docking for novel SARS-CoV-2 main protease non-covalent and covalent inhibitors

Antiviral therapeutics to treat SARS-CoV-2 are needed to diminish the morbidity of the ongoing COVID-19 pandemic. A well-precedented drug target is the main viral protease (MPro ), which is targeted by an approved drug and by several investigational drugs. Emerging viral resistance has made new inhibitor chemotypes more pressing. Adopting a structure-based approach, we docked 1.2 billion non-covalent lead-like molecules and a new library of 6.5 million electrophiles against the enzyme structure. From these, 29 non-covalent and 11 covalent inhibitors were identified in 37 series, the most potent having an IC50 of 29 and 20 μM, respectively. Several series were optimized, resulting in low micromolar inhibitors. Subsequent crystallography confirmed the docking predicted binding modes and may template further optimization. While the new chemotypes may aid further optimization of MPro inhibitors for SARS-CoV-2, the modest success rate also reveals weaknesses in our approach for challenging targets like MPro versus other targets where it has been more successful, and versus other structure-based techniques against MPro itself.

Structure-Based Discovery of Inhibitors of the SARS-CoV-2 Nsp14 N7-Methyltransferase

An under-explored target for SARS-CoV-2 is the S-adenosyl methionine (SAM)-dependent methyltransferase Nsp14, which methylates the N7-guanosine of viral RNA at the 5'-end, allowing the virus to evade host immune response. We sought new Nsp14 inhibitors with three large library docking strategies. First, up to 1.1 billion lead-like molecules were docked against the enzyme's SAM site, leading to three inhibitors with IC50 values from 6 to 50 μM. Second, docking a library of 16 million fragments revealed 9 new inhibitors with IC50 values from 12 to 341 μM. Third, docking a library of 25 million electrophiles to covalently modify Cys387 revealed 7 inhibitors with IC50 values from 3.5 to 39 μM. Overall, 32 inhibitors encompassing 11 chemotypes had IC50 values < 50 μM and 5 inhibitors in 4 chemotypes had IC50 values < 10 μM. These molecules are among the first non-SAM-like inhibitors of Nsp14, providing starting points for future optimization.

Modeling the expansion of virtual screening libraries

Recently, 'tangible' virtual libraries have made billions of molecules readily available. Prioritizing these molecules for synthesis and testing demands computational approaches, such as docking. Their success may depend on library diversity, their similarity to bio-like molecules and how receptor fit and artifacts change with library size. We compared a library of 3 million 'in-stock' molecules with billion-plus tangible libraries. The bias toward bio-like molecules in the tangible library decreases 19,000-fold versus those 'in-stock'. Similarly, thousands of high-ranking molecules, including experimental actives, from five ultra-large-library docking campaigns are also dissimilar to bio-like molecules. Meanwhile, better-fitting molecules are found as the library grows, with the score improving log-linearly with library size. Finally, as library size increases, so too do rare molecules that rank artifactually well. Although the nature of these artifacts changes from target to target, the expectation of their occurrence does not, and simple strategies can minimize their impact.

Synthesis, Computational Insights, and Evaluation of Novel Sigma Receptors Ligands

The development of diazabicyclo[4.3.0]nonane and 2,7-diazaspiro[3.5]nonane derivatives as sigma receptors (SRs) ligands is reported. The compounds were evaluated in S1R and S2R binding assays, and modeling studies were carried out to analyze the binding mode. The most notable compounds, 4b (AD186, K_iS1R = 2.7 nM, K_iS2R = 27 nM), 5b (AB21, K_iS1R = 13 nM, K_iS2R = 102 nM), and 8f (AB10, K_iS1R = 10 nM, K_iS2R = 165 nM), have been screened for analgesic effects in vivo, and their functional profile was determined through in vivo and in vitro models. Compounds 5b and 8f reached the maximum antiallodynic effect at 20 mg/kg. The selective S1R agonist PRE-084 completely reversed their action, indicating that the effects are entirely dependent on the S1R antagonism. Conversely, compound 4b sharing the 2,7-diazaspiro[3.5]nonane core as 5b was completely devoid of antiallodynic effect. Interestingly, compound 4b fully reversed the antiallodynic effect of BD-1063, indicating that 4b induces an S1R agonistic in vivo effect. The functional profiles were confirmed by the phenytoin assay. Our study might establish the importance of 2,7-diazaspiro[3.5]nonane core for the development of S1R compounds with specific agonist or antagonist profile and the role of the diazabicyclo[4.3.0]nonane in the development of novel SR ligands.

New avenues in artificial-intelligence-assisted drug discovery

Over the past decade, the amount of biomedical data available has grown at unprecedented rates. Increased automation technology and larger data volumes have encouraged the use of machine learning (ML) or artificial intelligence (AI) techniques for mining such data and extracting useful patterns. Because the identification of chemical entities with desired biological activity is a crucial task in drug discovery, AI technologies have the potential to accelerate this process and support decision making. In addition, the advent of deep learning (DL) has shown great promise in addressing diverse problems in drug discovery, such as de novo molecular design. Herein, we will appraise the current state-of-the-art in AI-assisted drug discovery, discussing the recent applications covering generative models for chemical structure generation, scoring functions to improve binding affinity and pose prediction, and molecular dynamics to assist in the parametrization, featurization and generalization tasks. Finally, we will discuss current hurdles and the strategies to overcome them, as well as potential future directions.

Reducing false positive rate of docking-based virtual screening by active learning

Machine learning-based scoring functions (MLSFs) have become a very favorable alternative to classical scoring functions because of their potential superior screening performance. However, the information of negative data used to construct MLSFs was rarely reported in the literature, and meanwhile the putative inactive molecules recorded in existing databases usually have obvious bias from active molecules. Here we proposed an easy-to-use method named AMLSF that combines active learning using negative molecular selection strategies with MLSF, which can iteratively improve the quality of inactive sets and thus reduce the false positive rate of virtual screening. We chose energy auxiliary terms learning as the MLSF and validated our method on eight targets in the diverse subset of DUD-E. For each target, we screened the IterBioScreen database by AMLSF and compared the screening results with those of the four control models. The results illustrate that the number of active molecules in the top 1000 molecules identified by AMLSF was significantly higher than those identified by the control models. In addition, the free energy calculation results for the top 10 molecules screened out by the AMLSF, null model and control models based on DUD-E also proved that more active molecules can be identified, and the false positive rate can be reduced by AMLSF.

Development of Optimal Virtual Screening Strategies to Identify Novel Toll-Like Receptor Ligands Using the DockBox Suite

Toll-like receptors (TLRs) represent attractive targets for developing modulators for the treatment of many pathologies, including inflammation, cancer, and autoimmune diseases. Here, we describe a protocol based on the DockBox package that enables to set up and perform structure-based virtual screening in order to increase the chance of identifying novel TLR ligands from chemical libraries.

Comprehensive Survey of Consensus Docking for High-Throughput Virtual Screening

The rapid advances of 3D techniques for the structural determination of proteins and the development of numerous computational methods and strategies have led to identifying highly active compounds in computer drug design. Molecular docking is a method widely used in high-throughput virtual screening campaigns to filter potential ligands targeted to proteins. A great variety of docking programs are currently available, which differ in the algorithms and approaches used to predict the binding mode and the affinity of the ligand. All programs heavily rely on scoring functions to accurately predict ligand binding affinity, and despite differences in performance, none of these docking programs is preferable to the others. To overcome this problem, consensus scoring methods improve the outcome of virtual screening by averaging the rank or score of individual molecules obtained from different docking programs. The successful application of consensus docking in high-throughput virtual screening highlights the need to optimize the predictive power of molecular docking methods.

Actives-Based Receptor Selection Strongly Increases the Success Rate in Structure-Based Drug Design and Leads to Identification of 22 Potent Cancer Inhibitors

Computer-aided drug design, an important component of the early stages of the drug discovery pipeline, routinely identifies large numbers of false positive hits that are subsequently confirmed to be experimentally inactive compounds. We have developed a methodology to improve true positive prediction rates in structure-based drug design and have successfully applied the protocol to twenty target systems and identified the top three performing conformers for each of the targets. Receptor performance was evaluated based on the area under the curve of the receiver operating characteristic curve for two independent sets of known actives. For a subset of five diverse cancer-related disease targets, we validated our approach through experimental testing of the top 50 compounds from a blind screening of a small molecule library containing hundreds of thousands of compounds. Our methods of receptor and compound selection resulted in the identification of 22 novel inhibitors in the low μM-nM range, with the most potent being an EGFR inhibitor with an IC₅₀ value of 7.96 nM. Additionally, for a subset of five independent target systems, we demonstrated the utility of Gaussian accelerated molecular dynamics to thoroughly explore a target system's potential energy surface and generate highly predictive receptor conformations.

Structure-Based Discovery of Negative Allosteric Modulators of the Metabotropic Glutamate Receptor 5

Recently determined structures of class C G protein-coupled receptors (GPCRs) revealed the location of allosteric binding sites and opened new opportunities for the discovery of novel modulators. In this work, molecular docking screens for allosteric modulators targeting the metabotropic glutamate receptor 5 (mGlu₅) were performed. The mGlu₅ receptor is activated by the main excitatory neurotransmitter of the nervous central system, L-glutamate, and mGlu₅ receptor activity can be allosterically modulated by negative or positive allosteric modulators. The mGlu₅ receptor is a promising target for the treatment of psychiatric and neurodegenerative diseases, and several allosteric modulators of this GPCR have been evaluated in clinical trials. Chemical libraries containing fragment- (1.6 million molecules) and lead-like (4.6 million molecules) compounds were docked to an allosteric binding site of mGlu₅ identified in X-ray crystal structures. Among the top-ranked compounds, 59 fragments and 59 lead-like compounds were selected for experimental evaluation. Of these, four fragment- and seven lead-like compounds were confirmed to bind to the allosteric site with affinities ranging from 0.43 to 8.6 μM, corresponding to a hit rate of 9%. The four compounds with the highest affinities were demonstrated to be negative allosteric modulators of mGlu₅ signaling in functional assays. The results demonstrate that virtual screens of fragment- and lead-like chemical libraries have complementary advantages and illustrate how access to high-resolution structures of GPCRs in complex with allosteric modulators can accelerate lead discovery.

Structure-based discovery of nonopioid analgesics acting through the α2A-adrenergic receptor

Because nonopioid analgesics are much sought after, we computationally docked more than 301 million virtual molecules against a validated pain target, the α2A-adrenergic receptor (α2AAR), seeking new α2AAR agonists chemotypes that lack the sedation conferred by known α2AAR drugs, such as dexmedetomidine. We identified 17 ligands with potencies as low as 12 nanomolar, many with partial agonism and preferential Gi and Go signaling. Experimental structures of α2AAR complexed with two of these agonists confirmed the docking predictions and templated further optimization. Several compounds, including the initial docking hit '9087 [mean effective concentration (EC50) of 52 nanomolar] and two analogs, '7075 and PS75 (EC50 4.1 and 4.8 nanomolar), exerted on-target analgesic activity in multiple in vivo pain models without sedation. These newly discovered agonists are interesting as therapeutic leads that lack the liabilities of opioids and the sedation of dexmedetomidine.

PyPLIF HIPPOS and Receptor Ensemble Docking Increase the Prediction Accuracy of the Structure-Based Virtual Screening Protocol Targeting Acetylcholinesterase

In this article, the upgrading process of the structure-based virtual screening (SBVS) protocol targeting acetylcholinesterase (AChE) previously published in 2017 is presented. The upgraded version of PyPLIF called PyPLIF HIPPOS and the receptor ensemble docking (RED) method using AutoDock Vina were employed to calculate the ensemble protein–ligand interaction fingerprints (ensPLIF) in a retrospective SBVS campaign targeting AChE. A machine learning technique called recursive partitioning and regression trees (RPART) was then used to optimize the prediction accuracy of the protocol by using the ensPLIF values as the descriptors. The best protocol resulting from this research outperformed the previously published SBVS protocol targeting AChE.