Virtual screening—an overview

Databases of ligand-binding pockets and protein-ligand interactions

Many research groups and institutions have created a variety of databases curating experimental and predicted data related to protein-ligand binding. The landscape of available databases is dynamic, with new databases emerging and established databases becoming defunct. Here, we review the current state of databases that contain binding pockets and protein-ligand binding interactions. We have compiled a list of such databases, fifty-three of which are currently available for use. We discuss variation in how binding pockets are defined and summarize pocket-finding methods. We organize the fifty-three databases into subgroups based on goals and contents, and describe standard use cases. We also illustrate that pockets within the same protein are characterized differently across different databases. Finally, we assess critical issues of sustainability, accessibility and redundancy.

Perspectives on current approaches to virtual screening in drug discovery

INTRODUCTION

For the past two decades, virtual screening (VS) has been an efficient hit finding approach for drug discovery. Today, billions of commercially accessible compounds are routinely screened, and many successful examples of VS have been reported. VS methods continue to evolve, including machine learning and physics-based methods.

AREAS COVERED

The authors examine recent examples of VS in drug discovery and discuss prospective hit finding results from the critical assessment of computational hit-finding experiments (CACHE) challenge. The authors also highlight the cost considerations and open-source options for conducting VS and examine chemical space coverage and library selections for VS.

EXPERT OPINION

The advancement of sophisticated VS approaches, including the use of machine learning techniques and increased computer resources as well as the ease of access to synthetically available chemical spaces, and commercial and open-source VS platforms allow for interrogating ultra-large libraries (ULL) of billions of molecules. An impressive number of prospective ULL VS campaigns have generated potent and structurally novel hits across many target classes. Nonetheless, many successful contemporary VS approaches still use considerably smaller focused libraries. This apparent dichotomy illustrates that VS is best conducted in a fit-for-purpose way choosing an appropriate chemical space. Better methods need to be developed to tackle more challenging targets.

Chemical space exploration with Molpher: Generating and assessing a glucocorticoid receptor ligand library

Computational exploration of chemical space is crucial in modern cheminformatics research for accelerating the discovery of new biologically active compounds. In this study, we present a detailed analysis of the chemical library of potential glucocorticoid receptor (GR) ligands generated by the molecular generator, Molpher. To generate the targeted GR library and construct the classification models, structures from the ChEMBL database as well as from the internal IMG library, which was experimentally screened for biological activity in the primary luciferase reporter cell assay, were utilized. The composition of the targeted GR ligand library was compared with a reference library that randomly samples chemical space. A random forest model was used to determine the biological activity of ligands, incorporating its applicability domain using conformal prediction. It was demonstrated that the GR library is significantly enriched with GR ligands compared to the random library. Furthermore, a prospective analysis demonstrated that Molpher successfully designed compounds, which were subsequently experimentally confirmed to be active on the GR. A collection of 34 potential new GR ligands was also identified. Moreover, an important contribution of this study is the establishment of a comprehensive workflow for evaluating computationally generated ligands, particularly those with potential activity against targets that are challenging to dock.

Discovery and in vitro Evaluation of Novel Serotonin 5-HT2A Receptor Ligands Identified Through Virtual Screening

The 5-HT2A receptor is a molecular target of high pharmacological importance. Ligands of this protein, particularly atypical antipsychotics, are useful in the treatment of numerous mental disorders, including schizophrenia and major depressive disorder. Structure-based virtual screening using a 5-HT2A receptor complex was performed to identify novel ligands for the 5-HT2A receptor, serving as potential antidepressants. From the Enamine screening library, containing over 4 million compounds, 48 molecules were selected for subsequent experimental validation. These compounds were tested against the 5-HT2A receptor in radioligand binding assays. From the tested batch, six molecules were identified as ligands of the main molecular target and were forwarded to a more detailed in vitro profiling. This included radioligand binding assays at 5-HT1A, 5-HT7, and D2 receptors and functional studies at 5-HT2A receptors. These compounds were confirmed to show a binding affinity for at least one of the targets tested in vitro. The success rate for the inactive template-based screening reached 17 %, while it was 9 % for the active template-based screening. Similarity and fragment analysis indicated the structural novelty of the identified compounds. Pharmacokinetics for these molecules was determined using in silico approaches.

In silico fragment-based discovery of CIB1-directed anti-tumor agents by FRASE-bot

Chemical probes are an indispensable tool for translating biological discoveries into new therapies, though are increasingly difficult to identify since novel therapeutic targets are often hard-to-drug proteins. We introduce FRASE-based hit-finding robot (FRASE-bot), to expedite drug discovery for unconventional therapeutic targets. FRASE-bot mines available 3D structures of ligand-protein complexes to create a database of FRAgments in Structural Environments (FRASE). The FRASE database can be screened to identify structural environments similar to those in the target protein and seed the target structure with relevant ligand fragments. A neural network model is used to retain fragments with the highest likelihood of being native binders. The seeded fragments then inform ultra-large-scale virtual screening of commercially available compounds. We apply FRASE-bot to identify ligands for Calcium and Integrin Binding protein 1 (CIB1), a promising drug target implicated in triple negative breast cancer. FRASE-based virtual screening identifies a small-molecule CIB1 ligand (with binding confirmed in a TR-FRET assay) showing specific cell-killing activity in CIB1-dependent cancer cells, but not in CIB1-depletion-insensitive cells.

A Comprehensive Survey on Deep Graph Representation Learning

Graph representation learning aims to effectively encode high-dimensional sparse graph-structured data into low-dimensional dense vectors, which is a fundamental task that has been widely studied in a range of fields, including machine learning and data mining. Classic graph embedding methods follow the basic idea that the embedding vectors of interconnected nodes in the graph can still maintain a relatively close distance, thereby preserving the structural information between the nodes in the graph. However, this is sub-optimal due to: (i) traditional methods have limited model capacity which limits the learning performance; (ii) existing techniques typically rely on unsupervised learning strategies and fail to couple with the latest learning paradigms; (iii) representation learning and downstream tasks are dependent on each other which should be jointly enhanced. With the remarkable success of deep learning, deep graph representation learning has shown great potential and advantages over shallow (traditional) methods, there exist a large number of deep graph representation learning techniques have been proposed in the past decade, especially graph neural networks. In this survey, we conduct a comprehensive survey on current deep graph representation learning algorithms by proposing a new taxonomy of existing state-of-the-art literature. Specifically, we systematically summarize the essential components of graph representation learning and categorize existing approaches by the ways of graph neural network architectures and the most recent advanced learning paradigms. Moreover, this survey also provides the practical and promising applications of deep graph representation learning. Last but not least, we state new perspectives and suggest challenging directions which deserve further investigations in the future.

Tackling assay interference associated with small molecules

Biochemical and cell-based assays are essential to discovering and optimizing efficacious and safe drugs, agrochemicals and cosmetics. However, false assay readouts stemming from colloidal aggregation, chemical reactivity, chelation, light signal attenuation and emission, membrane disruption, and other interference mechanisms remain a considerable challenge in screening synthetic compounds and natural products. To address assay interference, a range of powerful experimental approaches are available and in silico methods are now gaining traction. This Review begins with an overview of the scope and limitations of experimental approaches for tackling assay interference. It then focuses on theoretical methods, discusses strategies for their integration with experimental approaches, and provides recommendations for best practices. The Review closes with a summary of the critical facts and an outlook on potential future developments.

Identification and targeting of metastatic biomarkers for hepatocellular carcinoma therapeutics using small molecules library of curcumin analogues

The understanding of the molecular basis of complex diseases like hepatocellular carcinoma (HCC) needs large datasets of multiple genes and proteins involved in different phenomenon of its development. This study focuses on the molecular basis of HCC and the development of therapeutic strategies. We analyzed a dataset of 5475 genes (Homo sapiens) involved in HCC hallmarks, involving comprehensive data on multiple genes and frequently mutated genes. As HCC is characterized by metastasis, angiogenesis, and oxidative stress, exploration of genes associated with them has been targeted. Through gene ontology, functional characterization, and pathway enrichment analysis, we identified target proteins such as Lysyl oxidase, Survivin, Cofilin, and Cathepsin B. A library of curcumin analogs was used to target these proteins. Tetrahrydrocurcumin showed promising binding affinities for all four proteins, suggesting its potential as an inhibitor against these proteins for HCC therapy.

QSAR Study, Molecular Docking and Molecular Dynamic Simulation of Aurora Kinase Inhibitors Derived from Imidazo[4,5-b]pyridine Derivatives

Cancer is a serious threat to human life and social development and the use of scientific methods for cancer prevention and control is necessary. In this study, HQSAR, CoMFA, CoMSIA and TopomerCoMFA methods are used to establish models of 65 imidazo[4,5-b]pyridine derivatives to explore the quantitative structure-activity relationship between their anticancer activities and molecular conformations. The results show that the cross-validation coefficients q2 of HQSAR, CoMFA, CoMSIA and TopomerCoMFA are 0.892, 0.866, 0.877 and 0.905, respectively. The non-cross-validation coefficients r2 are 0.948, 0.983, 0.995 and 0.971, respectively. The externally validated complex correlation coefficients r2pred of external validation are 0.814, 0.829, 0.758 and 0.855, respectively. The PLS analysis verifies that the QSAR models have the highest prediction ability and stability. Based on these statistics, virtual screening based on R group is performed using the ZINC database by the Topomer search technology. Finally, 10 new compounds with higher activity are designed with the screened new fragments. In order to explore the binding modes and targets between ligands and protein receptors, these newly designed compounds are conjugated with macromolecular protein (PDB ID: 1MQ4) by molecular docking technology. Furthermore, to study the nature of the newly designed compound in dynamic states and the stability of the protein-ligand complex, molecular dynamics simulation is carried out for N3, N4, N5 and N7 docked with 1MQ4 protease structure for 50 ns. A free energy landscape is computed to search for the most stable conformation. These results prove the efficient and stability of the newly designed compounds. Finally, ADMET is used to predict the pharmacology and toxicity of the 10 designed drug molecules.

DFRscore: Deep Learning-Based Scoring of Synthetic Complexity with Drug-Focused Retrosynthetic Analysis for High-Throughput Virtual Screening

Recently emerging generative AI models enable us to produce a vast number of compounds for potential applications. While they can provide novel molecular structures, the synthetic feasibility of the generated molecules is often questioned. To address this issue, a few recent studies have attempted to use deep learning models to estimate the synthetic accessibility of many molecules rapidly. However, retrosynthetic analysis tools used to train the models rely on reaction templates automatically extracted from a large reaction database that are not domain-specific and may exhibit low chemical correctness. To overcome this limitation, we introduce DFRscore (Drug-Focused Retrosynthetic score), a deep learning-based approach for a more practical assessment of synthetic accessibility in drug discovery. The DFRscore model is trained exclusively on drug-focused reactions, providing a predicted number of minimally required synthetic steps for each compound. This approach enables practitioners to filter out compounds that do not meet their desired level of synthetic accessibility at an early stage of high-throughput virtual screening for accelerated drug discovery. The proposed strategy can be easily adapted to other domains by adjusting the synthesis planning setup of the reaction templates and starting materials.

AI is a viable alternative to high throughput screening: a 318-target study

High throughput screening (HTS) is routinely used to identify bioactive small molecules. This requires physical compounds, which limits coverage of accessible chemical space. Computational approaches combined with vast on-demand chemical libraries can access far greater chemical space, provided that the predictive accuracy is sufficient to identify useful molecules. Through the largest and most diverse virtual HTS campaign reported to date, comprising 318 individual projects, we demonstrate that our AtomNet® convolutional neural network successfully finds novel hits across every major therapeutic area and protein class. We address historical limitations of computational screening by demonstrating success for target proteins without known binders, high-quality X-ray crystal structures, or manual cherry-picking of compounds. We show that the molecules selected by the AtomNet® model are novel drug-like scaffolds rather than minor modifications to known bioactive compounds. Our empirical results suggest that computational methods can substantially replace HTS as the first step of small-molecule drug discovery.

Integration of fingerprint-based similarity searching and kernel-based partial least squares analysis to predict inhibitory activity against CSK, HER2, JAK1, JAK2, and JAK3

Fingerprint-based similarity searching is an important strategy for virtual screening in drug discovery. In the present study, we carried out a systematic virtual screening study, followed by the establishment of kernel-based partial least square (KPLS) analysis prediction models for five tyrosine kinase drug targets, C-terminal SRC kinase (CSK), human epidermal growth factor 2 (HER2), and Janus kinases 1, 2, and 3 (JAK1, JAK2, and JAK3), using a dataset of 3688 compounds. These kinases are important drug discovery targets, particularly as HER2 has been validated for the treatment of metastatic breast cancer, JAK inhibitors have been validated for the clinical management of arthritis and autoimmune diseases, and CSK has been found to play an important role in bone remodeling in arthritis. We conducted similarity screenings with the most active molecule for each target in the dataset as a query using eight (8) types of two-dimensional (2D) molecular fingerprints, comprising seven Hashed fingerprints, Linear, Dendritic, Radial, Pairwise, Triplet, Torsion, and MOLSPRINT2D, and one Structural keys fingerprint, MACCS. The top ranked 1% of compounds from each target's similarity screening results was used to set up kernel-based partial least square (KPLS) prediction models, with q2 values up to 0.8. The best KPLS model for each target was selected based on its predictive ability and boot strapping results and used for prediction. This integrated study approach combining similarity screening with KPLS analysis has a high potential to enhance the accuracy and efficiency of virtual screening and thus improve the drug discovery process.

Targeting P2Y14R protects against necroptosis of intestinal epithelial cells through PKA/CREB/RIPK1 axis in ulcerative colitis

Purinergic signaling plays a causal role in the pathogenesis of inflammatory bowel disease. Among purinoceptors, only P2Y14R is positively correlated with inflammatory score in mucosal biopsies of ulcerative colitis patients, nevertheless, the role of P2Y14R in ulcerative colitis remains unclear. Here, based on the over-expressions of P2Y14R in the intestinal epithelium of mice with experimental colitis, we find that male mice lacking P2Y14R in intestinal epithelial cells exhibit less intestinal injury induced by dextran sulfate sodium. Mechanistically, P2Y14R deletion limits the transcriptional activity of cAMP-response element binding protein through cAMP/PKA axis, which binds to the promoter of Ripk1, inhibiting necroptosis of intestinal epithelial cells. Furthermore, we design a hierarchical strategy combining virtual screening and chemical optimization to develop a P2Y14R antagonist HDL-16, which exhibits remarkable anti-colitis effects. Summarily, our study elucidates a previously unknown mechanism whereby P2Y14R participates in ulcerative colitis, providing a promising therapeutic target for inflammatory bowel disease.

Thompson Sampling─An Efficient Method for Searching Ultralarge Synthesis on Demand Databases

Over the last five years, virtual screening of ultralarge synthesis on-demand libraries has emerged as a powerful tool for hit identification in drug discovery programs. As these libraries have grown to tens of billions of molecules, we have reached a point where it is no longer cost-effective to screen every molecule virtually. To address these challenges, several groups have developed heuristic search methods to rapidly identify the best molecules on a virtual screen. This article describes the application of Thompson sampling (TS), an active learning approach that streamlines the virtual screening of large combinatorial libraries by performing a probabilistic search in the reagent space, thereby never requiring the full enumeration of the library. TS is a general technique that can be applied to various virtual screening modalities, including 2D and 3D similarity search, docking, and application of machine-learning models. In an illustrative example, we show that TS can identify more than half of the top 100 molecules from a docking-based virtual screen of 335 million molecules by evaluating 1% of the data set.

Integrating pharmacophore model and deep learning for activity prediction of molecules with BRCA1 gene

In this paper, we propose a novel approach for predicting the activity/inactivity of molecules with the BRCA1 gene by combining pharmacophore modeling and deep learning techniques. Initially, we generated 3D pharmacophore fingerprints using a pharmacophore model, which captures the essential features and spatial arrangements critical for biological activity. These fingerprints served as informative representations of the molecular structures. Next, we employed deep learning algorithms to train a predictive model using the generated pharmacophore fingerprints. The deep learning model was designed to learn complex patterns and relationships between the pharmacophore features and the corresponding activity/inactivity labels of the molecules. By utilizing this integrated approach, we aimed to enhance the accuracy and efficiency of activity prediction. To validate the effectiveness of our approach, we conducted experiments using a dataset of known molecules with BRCA1 gene activity/inactivity from diverse sources. Our results demonstrated promising predictive performance, indicating the successful integration of pharmacophore modeling and deep learning. Furthermore, we utilized the trained model to predict the activity/inactivity of unknown molecules extracted from the ChEMBL database. The predictions obtained from the ChEMBL database were assessed and compared against experimentally determined values to evaluate the reliability and generalizability of our model. Overall, our proposed approach showcased significant potential in accurately predicting the activity/inactivity of molecules with the BRCA1 gene, thus enabling the identification of potential candidates for further investigation in drug discovery and development processes.

Comparative Structure-Based Virtual Screening Utilizing Optimized AlphaFold Model Identifies Selective HDAC11 Inhibitor

HDAC11 is a class IV histone deacylase with no crystal structure reported so far. The catalytic domain of HDAC11 shares low sequence identity with other HDAC isoforms, which makes conventional homology modeling less reliable. AlphaFold is a machine learning approach that can predict the 3D structure of proteins with high accuracy even in absence of similar structures. However, the fact that AlphaFold models are predicted in the absence of small molecules and ions/cofactors complicates their utilization for drug design. Previously, we optimized an HDAC11 AlphaFold model by adding the catalytic zinc ion and minimization in the presence of reported HDAC11 inhibitors. In the current study, we implement a comparative structure-based virtual screening approach utilizing the previously optimized HDAC11 AlphaFold model to identify novel and selective HDAC11 inhibitors. The stepwise virtual screening approach was successful in identifying a hit that was subsequently tested using an in vitro enzymatic assay. The hit compound showed an IC50 value of 3.5 µM for HDAC11 and could selectively inhibit HDAC11 over other HDAC subtypes at 10 µM concentration. In addition, we carried out molecular dynamics simulations to further confirm the binding hypothesis obtained by the docking study. These results reinforce the previously presented AlphaFold optimization approach and confirm the applicability of AlphaFold models in the search for novel inhibitors for drug discovery.

Computer-Aided Drug Discovery and Design: Recent Advances and Future Prospects

Computer-aided drug discovery and design involve the use of information technologies to identify and develop, on a rational ground, chemical compounds that align a set of desired physicochemical and biological properties. In its most common form, it involves the identification and/or modification of an active scaffold (or the combination of known active scaffolds), although de novo drug design from scratch is also possible. Traditionally, the drug discovery and design processes have focused on the molecular determinants of the interactions between drug candidates and their known or intended pharmacological target(s). Nevertheless, in modern times, drug discovery and design are conceived as a particularly complex multiparameter optimization task, due to the complicated, often conflicting, property requirements.This chapter provides an updated overview of in silico approaches for identifying active scaffolds and guiding the subsequent optimization process. Recent groundbreaking advances in the field have also analyzed the integration of state-of-the-art machine learning approaches in every step of the drug discovery process (from prediction of target structure to customized molecular docking scoring functions), integration of multilevel omics data, and the use of a diversity of computational approaches to assist target validation and assess plausible binding pockets.

Data insights from a Moroccan phytochemical database (MPDB) derived from aromatic & medicinal plants

The geographical location of Morocco and the diversity of its topography ensure a high variability of climate conditions, ranging from humid to Saharan, and extending through subhumid, arid, and semi-arid stages. This variability offers a high floristic diversity, while the medical use of these phytochemicals has not been fully explored. Advanced computer-aided drug discovery utilizes chemical biology to accelerate the study of phytochemicals at the molecular level and discover novel therapeutic pathways. Currently, there is no online resource for phytochemicals in Morocco. Therefore, it is of interest to describe the Moroccan Phytochemicals Database (MPDB), accessible, featuring over 600 phytochemicals derived from journal articles and other reports. The web interface of the database, which is simple and easy to use, provides each phytochemical's reference, plant sources, 3D structures, and all related information. Furthermore, we provide direct links to commercially available analogs from Mcule. In addition, we provide the results of the first virtual screening against cardiovascular targets. We present these data to facilitate further exploration and exploitation of Morocco's rich phytochemical resources, and to contribute to the global understanding and application of these compounds in the medical and scientific communities.

Computational screening of coumarin derivatives as inhibitors of the NACHT domain of NLRP3 inflammasome for the treatment of Alzheimer's disease

The nucleotide-binding oligomerization domain (NOD)-like receptor (NLR), leucine-rich-repeat (LRR), and pyrin domain containing 3 (NLRP3) is one of the key players in neuroinflammation, which is a major pathological hallmark of Alzheimer's Disease (AD). Activated NLRP3 causes release of pro-inflammatory molecules that aggravate neurodegeneration. Thus, pharmacologically inhibiting the NLRP3 inflammasome has the potential to alleviate the inflammatory injury to the neurons. Coumarin is a multifunctional nucleus with potent anti-inflammatory properties and can be utilized to develop novel drugs for the treatment and management of AD. In the present study, we have explored the NLRP3-inhibitory activities of a library of coumarin derivatives through a computational drug discovery approach. Drug-like, PAINS free, and potentially BBB permeable compounds were screened out and subjected to molecular docking and in silico ADMET studies, resulting in three virtual hits, i.e. MolPort-050-872-358, MolPort-050-884-068, and MolPort-051-135-630. The hits exhibited better NLRP3-binding affinity than MCC950, a selective inhibitor of NLRP3. Further, molecular dynamics (MD) simulations, post-MD simulation analyses, and binding free energy calculations of the hits established their potential as promising virtual leads with a common coumarin scaffold for the inhibition of NLRP3 inflammasome.Communicated by Ramaswamy H. Sarma.

Structure-based virtual screening and molecular docking approaches to identify potential inhibitors against KIF2C to combat glioma

Glioma, a kind of malignant brain tumor, is extremely lethal. Kinesin family member 2C (KIF2C) was found to have an aberrant expression in several cancer types, including lung cancer and glioma. KIF2C may therefore be a useful therapeutic target for the treatment of glioma. In the current study, new drug candidates that may function as KIF2C enzyme inhibitors were discovered. MTi OpenScreen was used to carry out the structure-based virtual screening of an inbuilt drug library containing 150,000 compounds. These compounds belong to different classes, such as natural product-based compounds (NP-lib), purchasable approved drugs (Drugs-lib), and food constituents compound collection (FOOD-lib). Based on their binding affinities, a total of 84 compounds were further pushed to calculate ADMET properties. The compounds (16) meeting the ADMET cutoff ranges were then further docked to the receptor to find their plausible binding modes using the Glide tool's standard precision (SP) technique. The docking results were examined using the Glide gscore, and the best binding compounds (Rimacalib and Sarizotan) were chosen to test their stability with KIF2C protein through molecular dynamics (MD) simulation. Similarly, Principal Component Analysis and cross-correlation matrix were also examined. The MM/GBSA binding free energies showed a considerable energy contribution in the binding of hits with the KIF2C. Collectively, these findings strongly suggest the potential of the lead compounds to inhibit the biological function of KIF2C, emphasizing the need for further investigation in this area.Communicated by Ramaswamy H. Sarma.

Screening of potential antiplasmodial agents targeting cysteine protease-Falcipain 2: a computational pipeline

The spread of antimalarial drug resistance is a substantial challenge in achieving global malaria elimination. Consequently, the identification of novel therapeutic candidates is a global health priority. Malaria parasite necessitates hemoglobin degradation for its survival, which is mediated by Falcipain 2 (FP2), a promising antimalarial target. In particular, FP2 is a key enzyme in the erythrocytic stage of the parasite's life cycle. Here, we report the screening of approved drugs listed in DrugBank using a computational pipeline that includes drug-likeness, toxicity assessments, oral toxicity evaluation, oral bioavailability, docking analysis, maximum common substructure (MCS) and molecular dynamics (MD) Simulations analysis to identify capable FP2 inhibitors, which are hence potential antiplasmodial agents. A total of 45 drugs were identified, which have positive drug-likeness, no toxic features and good bioavailability. Among these, six drugs showed good binding affinity towards FP2 compared to E64, an epoxide known to inhibit FP2. Notably, two of them, Cefalotin and Cefoxitin, shared the highest MCS with E64, which suggests that they possess similar biological activity as E64. In an investigation using MD for 100 ns, Cefalotin and Cefoxitin showed adequate protein compactness as well as satisfactory complex stability. Overall, these computational approach findings can be applied for designing and developing specific inhibitors or new antimalarial agents for the treatment of malaria infections.Communicated by Ramaswamy H. Sarma.

Fake it until you make it? Generative de novo design and virtual screening of synthesizable molecules

Computational techniques, including virtual screening, de novo design, and generative models, play an increasing role in expediting DMTA cycles for modern molecular discovery. However, computationally proposed molecules must be synthetically feasible for laboratory testing. In this perspective, we offer a succinct introduction to the subject, and showcase typical workflows to integrate synthesis planning, synthesizability scoring, and molecule generation. Finally, we address limitations and opportunities for future research.

Integrating Reaction Schemes, Reagent Databases, and Virtual Libraries into Fragment-Based Design by Reinforcement Learning

Lead optimization supported by artificial intelligence (AI)-based generative models has become increasingly important in drug design. Success factors are reagent availability, novelty, and the optimization of multiple properties. Directed fragment-replacement is particularly attractive, as it mimics medicinal chemistry tactics. Here, we present variations of fragment-based reinforcement learning using an actor-critic model. Novel features include freezing fragments and using reagents as the fragment source. Splitting molecules according to reaction schemes improves synthesizability, while tuning network output probabilities allows us to balance novelty versus diversity. Combining fragment-based optimization with virtual library encodings allows the exploration of large chemical spaces with synthesizable ideas. Collectively, these enhancements influence design toward high-quality molecules with favorable profiles. A validation study using 15 pharmaceutically relevant targets reveals that novel structures are obtained for most cases, which are identical or related to independent validation sets for each target. Hence, these modifications significantly increase the value of fragment-based reinforcement learning for drug design. The code is available on GitHub: https://github.com/Sanofi-Public/IDD-papers-fragrl.

Cuprate superconducting materials above liquid nitrogen temperature from machine learning

The superconductivity of cuprates remains a challenging topic in condensed matter physics, and the search for materials that superconduct electricity above liquid nitrogen temperature and even at room temperature is of great significance for future applications. Nowadays, with the advent of artificial intelligence, research approaches based on data science have achieved excellent results in material exploration. We investigated machine learning (ML) models by employing separately the element symbolic descriptor atomic feature set 1 (AFS-1) and a prior physics knowledge descriptor atomic feature set 2 (AFS-2). An analysis of the manifold in the hidden layer of the deep neural network (DNN) showed that cuprates still offer the greatest potential as superconducting candidates. By calculating the SHapley Additive exPlanations (SHAP) value, it is evident that the covalent bond length and hole doping concentration emerge as the crucial factors influencing the superconducting critical temperature (T_c). These findings align with our current understanding of the subject, emphasizing the significance of these specific physical quantities. In order to improve the robustness and practicability of our model, two types of descriptors were used to train the DNN. We also proposed the idea of cost-sensitive learning, predicted the sample in another dataset, and designed a virtual high-throughput search workflow.

An Overview of the Potential Medicinal and Pharmaceutical Properties of Ru(II)/(III) Complexes

This review examines the existing knowledge about Ru(II)/(III) ion complexes with a potential application in medicine or pharmacy, which may offer greater potential in cancer chemotherapy than Pt(II) complexes, which are known to cause many side effects. Hence, much attention has been paid to research on cancer cell lines and clinical trials have been undertaken on ruthenium complexes. In addition to their antitumor activity, ruthenium complexes are under evaluation for other diseases, such as type 2 diabetes, Alzheimer's disease and HIV. Attempts are also being made to evaluate ruthenium complexes as potential photosensitizers with polypyridine ligands for use in cancer chemotherapy. The review also briefly examines theoretical approaches to studying the interactions of Ru(II)/Ru(III) complexes with biological receptors, which can facilitate the rational design of ruthenium-based drugs.

Recent Advances in Computer-Aided Structure-Based Drug Design on Ion Channels

Ion channels play important roles in fundamental biological processes, such as electric signaling in cells, muscle contraction, hormone secretion, and regulation of the immune response. Targeting ion channels with drugs represents a treatment option for neurological and cardiovascular diseases, muscular degradation disorders, and pathologies related to disturbed pain sensation. While there are more than 300 different ion channels in the human organism, drugs have been developed only for some of them and currently available drugs lack selectivity. Computational approaches are an indispensable tool for drug discovery and can speed up, especially, the early development stages of lead identification and optimization. The number of molecular structures of ion channels has considerably increased over the last ten years, providing new opportunities for structure-based drug development. This review summarizes important knowledge about ion channel classification, structure, mechanisms, and pathology with the main focus on recent developments in the field of computer-aided, structure-based drug design on ion channels. We highlight studies that link structural data with modeling and chemoinformatic approaches for the identification and characterization of new molecules targeting ion channels. These approaches hold great potential to advance research on ion channel drugs in the future.

Uni-Dock: GPU-Accelerated Docking Enables Ultralarge Virtual Screening

Molecular docking, a structure-based virtual screening method, is a reliable tool to enrich potential bioactive molecules from molecular databases. With the rapid expansion of compound library sizes, the speed of existing molecular docking programs becomes less than adequate to meet the demand for screening ultralarge libraries containing tens of millions or billions of molecules. Here, we propose Uni-Dock, a GPU-accelerated molecular docking program that supports various scoring functions including vina, vinardo, and ad4. Uni-Dock achieves more than 1000-fold speedup with high accuracy compared with the AutoDock Vina running in single CPU core, outperforming reported GPU-accelerated docking programs including AutoDock-GPU and Vina-GPU based on head-to-head experiments. Uni-Dock docks molecules in batches simultaneously using concurrent threads of each molecule. The data flow between GPU and CPU is optimized to eliminate CPU hotspots and maximize GPU utility. Additionally, Uni-Dock also supports hydrogen bond biased docking for all scoring functions and can be migrated to multiple GPUs of different architectures and manufacturers. We analyzed the improved performance of Uni-Dock on the CASF-2016 and DUD-E datasets and recommend three combinations of hyperparameters corresponding to different docking scenarios. To demonstrate Uni-Dock's capability on routinely screening ultralarge libraries, we performed hierarchical virtual screening experiments with Uni-Dock on the Enamine Diverse REAL druglike set containing 38.2 million molecules to a popular target KRAS G12D in 12 h using 100 NVIDIA V100 GPUs. To the best of our knowledge, Uni-Dock should be the fastest GPU-accelerated docking program to date.

Multitargeted Virtual Screening and Molecular Simulation of Natural Product-like Compounds against GSK3β, NMDA-Receptor, and BACE-1 for the Management of Alzheimer's Disease

The complexity of Alzheimer's disease (AD) and several side effects of currently available medication inclined us to search for a novel natural cure by targeting multiple key regulatory proteins. We initially virtually screened the natural product-like compounds against GSK3β, NMDA receptor, and BACE-1 and thereafter validated the best hit through molecular dynamics simulation (MDS). The results demonstrated that out of 2029 compounds, only 51 compounds exhibited better binding interactions than native ligands, with all three protein targets (NMDA, GSK3β, and BACE) considered multitarget inhibitors. Among them, F1094-0201 is the most potent inhibitor against multiple targets with binding energy -11.7, -10.6, and -12 kcal/mol, respectively. ADME-T analysis results showed that F1094-0201 was found to be suitable for CNS drug-likeness in addition to their other drug-likeness properties. The MDS results of RMSD, RMSF, Rg, SASA, SSE and residue interactions indicated the formation of a strong and stable association in the complex of ligands (F1094-0201) and proteins. These findings confirm the F1094-0201's ability to remain inside target proteins' binding pockets while forming a stable complex of protein-ligand. The free energies (MM/GBSA) of BACE-F1094-0201, GSK3β-F1094-0201, and NMDA-F1094-0201 complex formation were -73.78 ± 4.31 kcal mol^-1, -72.77 ± 3.43 kcal mol^-1, and -52.51 ± 2.85 kcal mol^-1, respectively. Amongst the target proteins, F1094-0201 have a more stable association with BACE, followed by NMDA and GSK3β. These attributes of F1094-0201 indicate it as a possible option for the management of pathophysiological pathways associated with AD.

The Impact of Supervised Learning Methods in Ultralarge High-Throughput Docking

Structure-based virtual screening methods are, nowadays, one of the key pillars of computational drug discovery. In recent years, a series of studies have reported docking-based virtual screening campaigns of large databases ranging from hundreds to thousands of millions compounds, further identifying novel hits after experimental validation. As these larg-scale efforts are not generally accessible, machine learning-based protocols have emerged to accelerate the identification of virtual hits within an ultralarge chemical space, reaching impressive reductions in computational time. Herein, we illustrate the motivation and the problem behind the screening of large databases, providing an overview of key concepts and essential applications of machine learning-accelerated protocols, specifically concerning supervised learning methods. We also discuss where the field stands with these novel developments, highlighting possible insights for future studies.

Protein-protein interfaces in molecular glue-induced ternary complexes: classification, characterization, and prediction

Molecular glues are a class of small molecules that stabilize the interactions between proteins. Naturally occurring molecular glues are present in many areas of biology where they serve as central regulators of signaling pathways. Importantly, several clinical compounds act as molecular glue degraders that stabilize interactions between E3 ubiquitin ligases and target proteins, leading to their degradation. Molecular glues hold promise as a new generation of therapeutic agents, including those molecular glue degraders that can redirect the protein degradation machinery in a precise way. However, rational discovery of molecular glues is difficult in part due to the lack of understanding of the protein-protein interactions they stabilize. In this review, we summarize the structures of known molecular glue-induced ternary complexes and the interface properties. Detailed analysis shows different mechanisms of ternary structure formation. Additionally, we also review computational approaches for predicting protein-protein interfaces and highlight the promises and challenges. This information will ultimately help inform future approaches for rational molecular glue discovery.

Machine Learning Scoring Functions for Drug Discovery from Experimental and Computer-Generated Protein-Ligand Structures: Towards Per-Target Scoring Functions

In recent years, machine learning has been proposed as a promising strategy to build accurate scoring functions for computational docking finalized to numerically empowered drug discovery. However, the latest studies have suggested that over-optimistic results had been reported due to the correlations present in the experimental databases used for training and testing. Here, we investigate the performance of an artificial neural network in binding affinity predictions, comparing results obtained using both experimental protein-ligand structures as well as larger sets of computer-generated structures created using commercial software. Interestingly, similar performances are obtained on both databases. We find a noticeable performance suppression when moving from random horizontal tests to vertical tests performed on target proteins not included in the training data. The possibility to train the network on relatively easily created computer-generated databases leads us to explore per-target scoring functions, trained and tested ad-hoc on complexes including only one target protein. Encouraging results are obtained, depending on the type of protein being addressed.

Structure-based virtual screening to identify potential lipase inhibitors to reduce lipid storage in Wolman disorder

Wolman disorder (WD) was first described in Iranian-Jewish (IJ) children, and it is caused by a deficiency of the lysosomal acid lipase (LAL). Newborns with WD are healthy and active at birth but soon develop severe malnutrition symptoms and often die before 1 year. In particular, spleens, livers, bone marrows, intestines, adrenal glands, and lymph nodes accumulate harmful amounts of lipids. G87V mutation in LIPA is responsible for Wolman disorder. Some reports suggest that δ-tocopherol can reduce lipid accumulation in cholesterol storage disorders. Hence, we used δ-tocopherol for the virtual screening process in this study. Initially, the lead compounds were docked with native and G87V mutant LIPA. Subsequently, the ADME and toxicity parameters for screened compounds were determined to ensure the safety profiles. Finally, the molecular dynamics simulations result indicated that dl-alpha-Tocopherol-13C3, a molecule obtained from the PubChem database, is identified as a potential and stable lead molecule that could be effective against the G87V mutant form of LIPA.

Machine Learning and Computational Chemistry for the Endocannabinoid System

Computational methods in medicinal chemistry facilitate drug discovery and design. In particular, machine learning methodologies have recently gained increasing attention. This chapter provides a structured overview of the current state of computational chemistry and its applications for the interrogation of the endocannabinoid system (ECS), highlighting methods in structure-based drug design, virtual screening, ligand-based quantitative structure-activity relationship (QSAR) modeling, and de novo molecular design. We emphasize emerging methods in machine learning and anticipate a forecast of future opportunities of computational medicinal chemistry for the ECS.

An Optimized Cloud Computing Method for Extracting Molecular Descriptors

Extracting molecular descriptors from chemical compounds is an essential preprocessing phase for developing accurate classification models. Supervised machine learning algorithms offer the capability to detect "hidden" patterns that may exist in a large dataset of compounds, which are represented by their molecular descriptors. Assuming that molecules with similar structure tend to share similar physicochemical properties, large chemical libraries can be screened by applying similarity sourcing techniques in order to detect potential bioactive compounds against a molecular target. However, the process of generating these compound features is time-consuming. Our proposed methodology not only employs cloud computing to accelerate the process of extracting molecular descriptors but also introduces an optimized approach to utilize the computational resources in the most efficient way.

A Hybrid Docking and Machine Learning Approach to Enhance the Performance of Virtual Screening Carried out on Protein-Protein Interfaces

The modulation of protein-protein interactions (PPIs) by small chemical compounds is challenging. PPIs play a critical role in most cellular processes and are involved in numerous disease pathways. As such, novel strategies that assist the design of PPI inhibitors are of major importance. We previously reported that the knowledge-based DLIGAND2 scoring tool was the best-rescoring function for improving receptor-based virtual screening (VS) performed with the Surflex docking engine applied to several PPI targets with experimentally known active and inactive compounds. Here, we extend our investigation by assessing the vs. potential of other types of scoring functions with an emphasis on docking-pose derived solvent accessible surface area (SASA) descriptors, with or without the use of machine learning (ML) classifiers. First, we explored rescoring strategies of Surflex-generated docking poses with five GOLD scoring functions (GoldScore, ChemScore, ASP, ChemPLP, ChemScore with Receptor Depth Scaling) and with consensus scoring. The top-ranked poses were post-processed to derive a set of protein and ligand SASA descriptors in the bound and unbound states, which were combined to derive descriptors of the docked protein-ligand complexes. Further, eight ML models (tree, bagged forest, random forest, Bayesian, support vector machine, logistic regression, neural network, and neural network with bagging) were trained using the derivatized SASA descriptors and validated on test sets. The results show that many SASA descriptors are better than Surflex and GOLD scoring functions in terms of overall performance and early recovery success on the used dataset. The ML models were superior to all scoring functions and rescoring approaches for most targets yielding up to a seven-fold increase in enrichment factors at 1% of the screened collections. In particular, the neural networks and random forest-based ML emerged as the best techniques for this PPI dataset, making them robust and attractive vs. tools for hit-finding efforts. The presented results suggest that exploring further docking-pose derived SASA descriptors could be valuable for structure-based virtual screening projects, and in the present case, to assist the rational design of small-molecule PPI inhibitors.

Concepts and applications of chemical fingerprint for hit and lead screening

Molecular fingerprints are used to represent chemical (structural, physicochemical, etc.) properties of large-scale chemical sets in a low computational cost way. They have a prominent role in transforming chemical data sets into consistent input formats (bit strings or numeric values) suitable for in silico approaches. In this review, we summarize and classify common and state-of-the-art fingerprints into eight different types (dictionary based, circular, topological, pharmacophore, protein-ligand interaction, shape based, reinforced, and multi). We also highlight applications of fingerprints in early drug research and development (R&D). Thus, this review provides a guide for the selection of appropriate fingerprints of compounds (or ligand-protein complexes) for use in drug R&D.

FNN Based-Virtual Screening Using 2D Pharmacophore Fingerprint for Activity Prediction in Drug Discovery

Drug discovery remains a hard field that faces from the beginning of its process to the end many difficulties and challenges in order to discover a new potential drug. The use of technology has helped a lot in achieving many goals at the lowest cost and in the shortest possible time. Machine learning methods have proven for many years their performance although their limitations in some cases. The use of deep learning for virtual screening in drug discovery allows to process efficiently the huge amount of data and gives more precise results. In this paper, we propose a procedure for virtual screening (VS) based on Feedforward Neural Network in order to predict the biological activity of a set of chemical compounds on a given receptor. we have proposed a distance interval and it divisions to describe the chemical compound by the 2D pharmacophore fingerprint. Our model was trained on a dataset of active and inactive chemical compounds on cyclin A kinase1 receptor (CDK1), a very important protein family which has a role in the regulation of the cell cycle and cancer development. The results have proven that the proposed model is efficient and comparable with some widely used machine learning methods in drug discovery.

Multi-Objective Drug Design Based on Graph-Fragment Molecular Representation and Deep Evolutionary Learning

Drug discovery is a challenging process with a huge molecular space to be explored and numerous pharmacological properties to be appropriately considered. Among various drug design protocols, fragment-based drug design is an effective way of constraining the search space and better utilizing biologically active compounds. Motivated by fragment-based drug search for a given protein target and the emergence of artificial intelligence (AI) approaches in this field, this work advances the field of in silico drug design by (1) integrating a graph fragmentation-based deep generative model with a deep evolutionary learning process for large-scale multi-objective molecular optimization, and (2) applying protein-ligand binding affinity scores together with other desired physicochemical properties as objectives. Our experiments show that the proposed method can generate novel molecules with improved property values and binding affinities.

Discovery of novel MIF inhibitors that attenuate microglial inflammatory activation by structures-based virtual screening and in vitro bioassays

Macrophage migration inhibitory factor (MIF) is a pluripotent pro-inflammatory cytokine and is related to acute and chronic inflammatory responses, immune disorders, tumors, and other diseases. In this study, an integrated virtual screening strategy and bioassays were used to search for potent MIF inhibitors. Twelve compounds with better bioactivity than the prototypical MIF-inhibitor ISO-1 (IC₅₀ = 14.41 μM) were identified by an in vitro enzymatic activity assay. Structural analysis revealed that these inhibitors have novel structural scaffolds. Compound 11 was then chosen for further characterization in vitro, and it exhibited marked anti-inflammatory efficacy in LPS-activated BV-2 microglial cells by suppressing the activation of nuclear factor kappa B (NF-κB) and mitogen-activated protein kinases (MAPKs). Our findings suggest that MIF may be involved in the regulation of microglial inflammatory activation and that small-molecule MIF inhibitors may serve as promising therapeutic agents for neuroinflammatory diseases.

Insights into the Pharmacological Effects of Flavonoids: The Systematic Review of Computer Modeling

Computer modeling is a method that is widely used in scientific investigations to predict the biological activity, toxicity, pharmacokinetics, and synthesis strategy of compounds based on the structure of the molecule. This work is a systematic review of articles performed in accordance with the recommendations of PRISMA and contains information on computer modeling of the interaction of classical flavonoids with different biological targets. The review of used computational approaches is presented. Furthermore, the affinities of flavonoids to different targets that are associated with the infection, cardiovascular, and oncological diseases are discussed. Additionally, the methodology of bias risks in molecular docking research based on principles of evidentiary medicine was suggested and discussed. Based on this data, the most active groups of flavonoids and lead compounds for different targets were determined. It was concluded that flavonoids are a promising object for drug development and further research of pharmacology by in vitro, ex vivo, and in vivo models is required.

Comparative Analyses of Medicinal Chemistry and Cheminformatics Filters with Accessible Implementation in Konstanz Information Miner (KNIME)

High-throughput virtual screening (HTVS) is, in conjunction with rapid advances in computer hardware, becoming a staple in drug design research campaigns and cheminformatics. In this context, virtual compound library design becomes crucial as it generally constitutes the first step where quality filtered databases are essential for the efficient downstream research. Therefore, multiple filters for compound library design were devised and reported in the scientific literature. We collected the most common filters in medicinal chemistry (PAINS, REOS, Aggregators, van de Waterbeemd, Oprea, Fichert, Ghose, Mozzicconacci, Muegge, Egan, Murcko, Veber, Ro3, Ro4, and Ro5) to facilitate their open access use and compared them. Then, we implemented these filters in the open platform Konstanz Information Miner (KNIME) as a freely accessible and simple workflow compatible with small or large compound databases for the benefit of the readers and for the help in the early drug design steps.

Artificial intelligence in virtual screening: models versus experiments

A typical drug discovery project involves identifying active compounds with significant binding potential for selected disease-specific targets. Experimental high-throughput screening (HTS) is a traditional approach to drug discovery, but is expensive and time-consuming when dealing with huge chemical libraries with billions of compounds. The search space can be narrowed down with the use of reliable computational screening approaches. In this review, we focus on various machine-learning (ML) and deep-learning (DL)-based scoring functions developed for solving classification and ranking problems in drug discovery. We highlight studies in which ML and DL models were successfully deployed to identify lead compounds for which the experimental validations are available from bioassay studies.

Machine Learning and Its Applications for Protozoal Pathogens and Protozoal Infectious Diseases

In recent years, massive attention has been attracted to the development and application of machine learning (ML) in the field of infectious diseases, not only serving as a catalyst for academic studies but also as a key means of detecting pathogenic microorganisms, implementing public health surveillance, exploring host-pathogen interactions, discovering drug and vaccine candidates, and so forth. These applications also include the management of infectious diseases caused by protozoal pathogens, such as Plasmodium, Trypanosoma, Toxoplasma, Cryptosporidium, and Giardia, a class of fatal or life-threatening causative agents capable of infecting humans and a wide range of animals. With the reduction of computational cost, availability of effective ML algorithms, popularization of ML tools, and accumulation of high-throughput data, it is possible to implement the integration of ML applications into increasing scientific research related to protozoal infection. Here, we will present a brief overview of important concepts in ML serving as background knowledge, with a focus on basic workflows, popular algorithms (e.g., support vector machine, random forest, and neural networks), feature extraction and selection, and model evaluation metrics. We will then review current ML applications and major advances concerning protozoal pathogens and protozoal infectious diseases through combination with correlative biology expertise and provide forward-looking insights for perspectives and opportunities in future advances in ML techniques in this field.

Personalized Liver Cancer Risk Prediction Using Big Data Analytics Techniques with Image Processing Segmentation

A technology known as data analytics is a massively parallel processing approach that may be used to forecast a wide range of illnesses. Many scientific research methodologies have the problem of requiring a significant amount of time and processing effort, which has a negative impact on the overall performance of the system. Virtual screening (VS) is a drug discovery approach that makes use of big data techniques and is based on the concept of virtual screening. This approach is utilised for the development of novel drugs, and it is a time-consuming procedure that includes the docking of ligands in several databases in order to build the protein receptor. The proposed work is divided into two modules: image processing-based cancer segmentation and analysis using extracted features using big data analytics, and cancer segmentation and analysis using extracted features using image processing. This statistical approach is critical in the development of new drugs for the treatment of liver cancer. Machine learning methods were utilised in the prediction of liver cancer, including the MapReduce and Mahout algorithms, which were used to prefilter the set of ligand filaments before they were used in the prediction of liver cancer. This work proposes the SMRF algorithm, an improved scalable random forest algorithm built on the MapReduce foundation. Using a computer cluster or cloud computing environment, this new method categorises massive datasets. With SMRF, small amounts of data are processed and optimised over a large number of computers, allowing for the highest possible throughput. When compared to the standard random forest method, the testing findings reveal that the SMRF algorithm exhibits the same level of accuracy deterioration but exhibits superior overall performance. The accuracy range of 80 percent using the performance metrics analysis is included in the actual formulation of the medicine that is utilised for liver cancer prediction in this study.

Redox active or thiol reactive? Optimization of rapid screens to identify less evident nuisance compounds

Compounds that exhibit assay interference or undesirable mechanisms of bioactivity are routinely encountered in assays at various stages of drug discovery. We observed that assays for the investigation of thiol-reactive and redox-active compounds have not been collected in a comprehensive review. Here, we review these assays and subject them to experimental optimization to improve their reliability. We demonstrate the usefulness of our assay cascade by assaying a library of bioactive compounds, chemical probes, and a set of approved drugs. These high-throughput assays should complement the array of wet-lab and in silico assays during the initial stages of hit discovery campaigns to pursue only hit compounds with tractable mechanisms of action. Teaser: We provide an overview of assays to detect redox active and thiol reactive compounds and the robust protocols for identification of nuisance compounds during early stages of drug discovery programs.

Application of deep metric learning to molecular graph similarity

Graph based methods are increasingly important in chemistry and drug discovery, with applications ranging from QSAR to molecular generation. Combining graph neural networks and deep metric learning concepts, we expose a framework for quantifying molecular graph similarity based on distance between learned embeddings separate from any endpoint. Using a minimal definition of similarity, and data from the ZINC database of public compounds, this work demonstrate the properties of the embedding and its suitability for a range of applications, among them a novel reconstruction loss method for training deep molecular auto-encoders. Finally, we compare the applications of the embedding to standard practices, with a focus on known failure points and edge cases; concluding that our approach can be used in conjunction to existing methods.

On the Value of Using 3D Shape and Electrostatic Similarities in Deep Generative Methods

Multiparameter optimization, the heart of drug design, is still an open challenge. Thus, improved methods for automated compound design with multiple controlled properties are desired. Here, we present a significant extension to our previously described fragment-based reinforcement learning method (DeepFMPO) for the generation of novel molecules with optimal properties. As before, the generative process outputs optimized molecules similar to the input structures, now with the improved feature of replacing parts of these molecules with fragments of similar three-dimensional (3D) shape and electrostatics. We developed and benchmarked a new python package, ESP-Sim, for the comparison of the electrostatic potential and the molecular shape, allowing the calculation of high-quality partial charges (e.g., RESP with B3LYP/6-31G**) obtained using the quantum chemistry program Psi4. By performing comparisons of 3D fragments, we can simulate 3D properties while overcoming the notoriously difficult step of accurately describing bioactive conformations. The new improved generative (DeepFMPO v3D) method is demonstrated with a scaffold-hopping exercise identifying CDK2 bioisosteres. The code is open-source and freely available.

Discovery of Small-Molecule Inhibitors of the PD-1/PD-L1 Axis That Promote PD-L1 Internalization and Degradation

Several monoclonal antibodies targeting the programmed cell death-1/programmed cell death-ligand 1 (PD-1/PD-L1) pathway have been used successfully in anticancer immunotherapy. Inherent limitations of antibody-based therapies remain, however, and alternative small-molecule inhibitors that can block the PD-1/PD-L1 axis are urgent needed. Herein, we report the discovery of compound 17 as a bifunctional inhibitor of PD-1/PD-L1 interactions. 17 inhibits PD-1/PD-L1 interactions and promotes dimerization, internalization, and degradation of PD-L1. 17 promotes cell-surface PD-L1 internalized into the cytosol and induces the degradation of PD-L1 in tumor cells through a lysosome-dependent pathway. Furthermore, 17 suppresses tumor growth in vivo by activating antitumor immunity. These results demonstrate that 17 targets the PD-1/PD-L1 axis and induces PD-L1 degradation.