BindingDB in 2015: A public database for medicinal chemistry, computational chemistry and systems pharmacology

The future of metronomic chemotherapy: experimental and computational approaches of drug repurposing

Metronomic chemotherapy (MC), long-term continuous administration of anticancer drugs, is gaining attention as an alternative to the traditional maximum tolerated dose (MTD) chemotherapy. By combining MC with other treatments, the therapeutic efficacy is enhanced while minimizing toxicity. MC employs multiple mechanisms, making it a versatile approach against various cancers. However, drug resistance limits the long-term effectiveness of MC, necessitating ongoing development of anticancer drugs. Traditional drug discovery is lengthy and costly due to processes like target protein identification, virtual screening, lead optimization, and safety and efficacy evaluations. Drug repurposing (DR), which screens FDA-approved drugs for new uses, is emerging as a cost-effective alternative. Both experimental and computational methods, such as protein binding assays, in vitro cytotoxicity tests, structure-based screening, and several types of association analyses (Similarity-Based, Network-Based, and Target Gene), along with retrospective clinical analyses, are employed for virtual screening. This review covers the mechanisms of MC, its application in various cancers, DR strategies, examples of repurposed drugs, and the associated challenges and future directions.

Research progress on Drug-Target Interactions in the last five years

The identification of Drug-Target Interaction (DTI) is an important step in drug discovery and drug repositioning, and has high application value in multiple fields such as drug discovery, drug repositioning, and repurposing. However, the high cost of experimental validation limits its identification. In contrast, computation-based approaches are both economical and efficient. This review first synthesizes existing chemical genomic approaches, provides a comprehensive summary of prevalent databases for predicting DTIs, and categorizes the feature encodings from recent years. This is followed by an overview and brief description of the methods currently in use for predicting DTIs. The strengths and weaknesses of newly proposed prediction methods in the last five years (2020-2024), including those based on network representation learning and graph neural networks, are then discussed in detail, evaluating the performance of the different methods on a wide range of datasets. Finally, this review explores potential directions for future DTI research, emphasizing how to improve prediction accuracy and efficiency by combining big data and emerging computing technologies.

Novel nano-sized N-Thiazolylpyridylamines targeting CDK2: Design, divergent synthesis, conformational studies, and multifaceted In silico analysis

This study involves the design, divergent synthesis, conformational and structural analysis, target prediction, and molecular docking simulations of novel nano N-thiazolylpyridylamines 2-7 and 10 as potential cyclin-dependent kinase 2 (CDK2) inhibitors. Using a divergent synthesis approach, the compounds were designed with structural variation and optimization in mind. The conformational and structural properties were explored through various spectroscopic techniques, confirming the structure, stability, and preferred conformations. Additionally, nanocrystalline characterization, including X-ray diffraction analysis, revealed the nanoscale structural features of the synthesized molecules. Most compounds exhibited a crystalline nature with crystallite sizes ranging from 10.75 to 57.77 nm, which is crucial for improving cellular uptake and anticancer efficacy. Biological testing was performed to evaluate the cytotoxicity of compounds 2-7 and 10 against cancer cell lines, including HepG2, MCF-7, and HCT-116. Compound 5 exhibited significant cytotoxicity with IC50 values of 10.9 ± 0.5 μM, 6.98 ± 0.3 μM, and 6.3 ± 0.2 μM against MCF-7, HePG2, and HCT116, respectively. Other compounds demonstrated varied activities, with compounds 4, 6, and 10 showing moderate activity against the MCF-7 cell line. Computational techniques suggested a strong probability of these compounds targeting CDK2, with molecular docking and dynamics used to predict their binding mechanisms. These findings suggest that N-thiazolylpyridylamines may serve as new anticancer agents for further lead optimization.

Overview and limitations of database in global traditional medicines: A narrative review

The study of traditional medicine has garnered significant interest, resulting in various research areas including chemical composition analysis, pharmacological research, clinical application, and quality control. The abundance of available data has made databases increasingly essential for researchers to manage the vast amount of information and explore new drugs. In this article we provide a comprehensive overview and summary of 182 databases that are relevant to traditional medicine research, including 73 databases for chemical component analysis, 70 for pharmacology research, and 39 for clinical application and quality control from published literature (2000-2023). The review categorizes the databases by functionality, offering detailed information on websites and capacities to facilitate easier access. Moreover, this article outlines the primary function of each database, supplemented by case studies to aid in database selection. A practical test was conducted on 68 frequently used databases using keywords and functionalities, resulting in the identification of highlighted databases. This review serves as a reference for traditional medicine researchers to choose appropriate databases and also provides insights and considerations for the function and content design of future databases.

HEDDI-Net: heterogeneous network embedding for drug-disease association prediction and drug repurposing, with application to Alzheimer's disease

BACKGROUND

The traditional process of developing new drugs is time-consuming and often unsuccessful, making drug repurposing an appealing alternative due to its speed and safety. Graph neural networks (GCNs) have emerged as a leading approach for predicting drug-disease associations by integrating drug and disease-related networks with advanced deep learning algorithms. However, GCNs generally infer association probabilities only for existing drugs and diseases, requiring network re-establishment and retraining for novel entities. Additionally, these methods often struggle with sparse networks and fail to elucidate the biological mechanisms underlying newly predicted drugs.

METHODS

To address the limitations of traditional methods, we developed HEDDI-Net, a heterogeneous embedding architecture designed to accurately detect drug-disease associations while preserving the interpretability of biological mechanisms. HEDDI-Net integrates graph and shallow learning techniques to extract representative diseases and proteins, respectively. These representative diseases and proteins are used to embed the input features, which are then utilized in a multilayer perceptron for predicting drug-disease associations.

RESULTS

In experiments, HEDDI-Net achieves areas under the receiver operating characteristic curve of over 0.98, outperforming state-of-the-art methods. Rigorous recovery analyses reveal a median recovery rate of 73% for the top 100 diseases, demonstrating its efficacy in identifying novel target diseases for existing drugs, known as drug repurposing. A case study on Alzheimer's disease highlighted the model's practical applicability and interpretability, identifying potential drug candidates like Baclofen, Fluoxetine, Pentoxifylline and Phenytoin. Notably, over 40% of the predicted candidates in the clusters of commonly prescribed clinical drugs Donepezil and Galantamine had been tested in clinical trials, validating the model's predictive accuracy and practical relevance.

CONCLUSIONS

HEDDI-NET represents a significant advancement by allowing direct application to new diseases and drugs without the need for retraining, a limitation of most GCN-based methods. Furthermore, HEDDI-Net provides detailed affinity patterns with representative proteins for predicted candidate drugs, facilitating an understanding of their physiological effects. This capability also supports the design and testing of alternative drugs that are similar to existing medications, enhancing the reliability and interpretability of potential repurposed drugs. The case study on Alzheimer's disease further underscores HEDDI-Net's ability to predict promising drugs and its applicability in drug repurposing.

APBIO: bioactive profiling of air pollutants through inferred bioactivity signatures and prediction of novel target interactions

More sophisticated representations of compounds attempt to incorporate not only information on the structure and physicochemical properties of molecules, but also knowledge about their biological traits, leading to the so-called bioactivity profile. The bioactive profiling of air pollutants is challenging and crucial, as their biological activity and toxicological effects have not been deeply investigated yet, and further exploration could shed light on the impact of air pollution on complex disorders. Therefore, a biological signature that simultaneously captures the chemistry and the biology of small molecules may be beneficial in predicting the behaviour of such ligands towards a protein target. Moreover, the interactivity between biological entities can be represented through combined feature vectors that can be given as input to a machine learning (ML) model to capture the underlying interaction. To this end, we propose a chemogenomic approach, called Air Pollutant Bioactivity (APBIO), which integrates compound bioactivity signatures and target sequence descriptors to train ML classifiers subsequently used to predict potential compound-target interactions (CTIs). We report the performances of the proposed methodology and, via external validation sets, demonstrate its outperformance compared to existing molecular representations in terms of model generalizability. We have also developed a publicly available Streamlit application for APBIO at ap-bio.streamlit.app, allowing users to predict associations between investigated compounds and protein targets.Scientific contributionWe derived ex novo bioactivity signatures for air pollutant molecules to capture their biological behaviour and associations with protein targets. The proposed chemogenomic methodology enables the prediction of novel CTIs for known or similar compounds and targets through well-established and efficient ML models, deepening our insight into the molecular interactions and mechanisms that may have a deleterious impact on human biological systems.

DO-GMA: An End-to-End Drug-Target Interaction Identification Framework with a Depthwise Overparameterized Convolutional Network and the Gated Multihead Attention Mechanism

Identification of potential drug-target interactions (DTIs) is a crucial step in drug discovery and repurposing. Although deep learning effectively deciphers DTIs, most deep learning-based methods represent drug features from only a single perspective. Moreover, the fusion method of drug and protein features needs further refinement. To address the above two problems, in this study, we develop a novel end-to-end framework named DO-GMA for potential DTI identification by incorporating Depthwise Overparameterized convolutional neural network and the Gated Multihead Attention mechanism with shared-learned queries and bilinear model concatenation. DO-GMA first designs a depthwise overparameterized convolutional neural network to learn drug representations from their SMILES strings and protein representations from their amino acid sequences. Next, it extracts drug representations from their 2D molecular graphs through a graph convolutional network. Subsequently, it fuses drug and protein features by combining the gated attention mechanism and the multihead attention mechanism with shared-learned queries and bilinear model concatenation. Finally, it takes the fused drug-target features as inputs and builds a multilayer perceptron to classify unlabeled drug-target pairs (DTPs). DO-GMA was benchmarked against six newest DTI prediction methods (CPI-GNN, BACPI, CPGL, DrugBAN, BINDTI, and FOTF-CPI) under four different experimental settings on four DTI data sets (i.e., DrugBank, BioSNAP, C.elegans, and BindingDB). The results show that DO-GMA significantly outperformed the above six methods based on AUC, AUPR, accuracy, F1-score, and MCC. An ablation study, robust statistical analysis, sensitivity analysis of parameters, visualization of the fused features, computational cost analysis, and case analysis further validated the powerful DTI identification performance of DO-GMA. In addition, DO-GMA predicted that two drug-protein pairs (i.e., DB00568 and P06276, and DB09118 and Q9UQD0) could be interacting. DO-GMA is freely available at https://github.com/plhhnu/DO-GMA.

Artificial intelligence streamlines scientific discovery of drug-target interactions

Drug discovery is a complicated process through which new therapeutics are identified to prevent and treat specific diseases. Identification of drug-target interactions (DTIs) stands as a pivotal aspect within the realm of drug discovery and development. The traditional process of drug discovery, especially identification of DTIs, is marked by its high costs of experimental assays and low success rates. Computational methods have emerged as indispensable tools, especially those employing artificial intelligence (AI) methods, which could streamline the process, thereby reducing costs and time consumption and potentially increasing success rates. In this review, we focus on the application of AI techniques in DTI prediction. Specifically, we commence with a comprehensive overview of drug discovery and development, along with systematic prediction and validation of DTIs. We proceed to highlight the prominent databases and toolkits used in developing AI methods for DTI prediction, as well as with methodologies for evaluating their efficacy. We further extend the exploration into three primary types of state-of-the-art AI methods used in DTI prediction, including classical machine learning, deep learning and network-based methods. Finally, we summarize the key findings and outline the current challenges and future directions that AI methods face in scientific drug discovery and development.

Deep learning methods for proteome-scale interaction prediction

Proteome-scale interaction prediction is essential for understanding protein functions and disease mechanisms. Traditional experimental methods are often limited by scale and complexity, driving the need for computational approaches. Deep learning has emerged as a powerful tool, enabling high-throughput, accurate predictions of protein interactions. This review highlights recent advances in deep learning methods for protein-protein and protein-ligand interaction screening, along with datasets used for model training. Despite the progress with deep learning, challenges such as data quality and validation biases remain. We also discuss the increasing importance of integrating structural information to enhance prediction accuracy and how structure-based deep learning approaches can help overcome current limitations, ultimately advancing biological research and drug discovery.

General structure-activity relationship models for the inhibitors of Adenosine receptors: A machine learning approach

Adenosine receptors (A1, A2a, A2b, A3) play critical roles in cellular signaling and are implicated in various physiological and pathological processes, including inflammations and cancer. The main aim of this research was to investigate structure-activity relationships (SAR) to derive models that describe the selectivity and activity of inhibitors targeting Adenosine receptors. Structural information for 16,312 inhibitors was collected from BindingDB and analyzed using machine learning methods. 450 molecular descriptors were calculated for each molecule and compounds were classified based on their activity levels and therapeutic targets. The variable importance in projection (VIP) algorithm identified key discriminating features. Classification models were built using supervised Kohonen networks (SKN) and counter-propagation artificial neural networks (CPANN) algorithms. Model validity was assessed via cross-validation, applicability domain analysis, and test sets. These models were then used to screen a random subset of 2 million molecules from the ZINC database. Three descriptors-hydrophilic factor (Hy), ratio of multiple path count over path count (PCR), and asphericity (ASP)-were identified as critical for discriminating active and inactive inhibitors. SKN models exhibited high sensitivity (0.88-0.99) and yielded an average area under the curve (AUC) of 0.922 for virtual screening. This study aimed to enhance the development of highly selective Adenosine receptor ligands for diverse therapeutic applications by identifying critical molecular features specific to each isoform.

DeepDrug as an expert guided and AI driven drug repurposing methodology for selecting the lead combination of drugs for Alzheimer's disease

Alzheimer's Disease (AD) significantly aggravates human dignity and quality of life. While newly approved amyloid immunotherapy has been reported, effective AD drugs remain to be identified. Here, we propose a novel AI-driven drug-repurposing method, DeepDrug, to identify a lead combination of approved drugs to treat AD patients. DeepDrug advances drug-repurposing methodology in four aspects. Firstly, it incorporates expert knowledge to extend candidate targets to include long genes, immunological and aging pathways, and somatic mutation markers that are associated with AD. Secondly, it incorporates a signed directed heterogeneous biomedical graph encompassing a rich set of nodes and edges, and node/edge weighting to capture crucial pathways associated with AD. Thirdly, it encodes the weighted biomedical graph through a Graph Neural Network into a new embedding space to capture the granular relationships across different nodes. Fourthly, it systematically selects the high-order drug combinations via diminishing return-based thresholds. A five-drug lead combination, consisting of Tofacitinib, Niraparib, Baricitinib, Empagliflozin, and Doxercalciferol, has been selected from the top drug candidates based on DeepDrug scores to achieve the maximum synergistic effect. These five drugs target neuroinflammation, mitochondrial dysfunction, and glucose metabolism, which are all related to AD pathology. DeepDrug offers a novel AI-and-big-data, expert-guided mechanism for new drug combination discovery and drug-repurposing across AD and other neuro-degenerative diseases, with immediate clinical applications.

TopMT-GAN: a 3D topology-driven generative model for efficient and diverse structure-based ligand design

Recent advancements in 3D structure-based molecular generative models have shown promise in expediting the hit discovery process in drug design. Despite their potential, efficiently generating a focused library of candidate molecules that exhibit both effective interactions and structural diversity at a large scale remains a significant challenge. Moreover, current studies often lack comprehensive comparisons to high-throughput virtual screening methods, resulting in insufficient evaluation of their effectiveness. In this study, we introduce Topology Molecular Type assignment (TopMT-GAN), a novel approach using Generative Adversarial Networks (GANs) for direct structure-based design. TopMT-GAN employs a two-step strategy: constructing 3D molecular topologies within a protein pocket with one GAN, followed by atom and bond type assignment with a second GAN. This integrated approach enables TopMT-GAN to efficiently generate diverse and potent ligands with precise 3D poses for specific protein pockets. When tested on five diverse protein pockets, TopMT-GAN exhibits promising and robust performance, demonstrating a potential enrichment of up to 46 000 fold compared to traditional high-throughput virtual screening methods. This highlights its potential as a powerful tool in early-stage drug discovery, such as hit and lead generation.

canSAR 2024-an update to the public drug discovery knowledgebase

canSAR (https://cansar.ai) continues to serve as the largest publicly available platform for cancer-focused drug discovery and translational research. It integrates multidisciplinary data from disparate and otherwise siloed public data sources as well as data curated uniquely for canSAR. In addition, canSAR deploys a suite of curation and standardization tools together with AI algorithms to generate new knowledge from these integrated data to inform hypothesis generation. Here we report the latest updates to canSAR. As well as increasing available data, we provide enhancements to our algorithms to improve the offering to the user. Notably, our enhancements include a revised ligandability classifier leveraging Positive Unlabeled Learning that finds twice as many ligandable opportunities across the pocketome, and our revised chemical standardization pipeline and hierarchy better enables the aggregation of structurally related molecular records.

ECBD: European chemical biology database

The European Chemical Biology Database (ECBD, https://ecbd.eu) serves as the central repository for data generated by the EU-OPENSCREEN research infrastructure consortium. It is developed according to FAIR principles, which emphasize findability, accessibility, interoperability and reusability of data. This data is made available to the scientific community following open access principles. The ECBD stores both positive and negative results from the entire chemical biology project pipeline, including data from primary or counter-screening assays. The assays utilize a defined and diverse library of over 107 000 compounds, the annotations of which are continuously enriched by external user supported screening projects and by internal EU-OPENSCREEN bioprofiling efforts. These compounds were screened in 89 currently deposited datasets (assays), with 48 already being publicly accessible, while the remaining will be published after a publication embargo period of up to 3 years. Together these datasets encompass ∼4.3 million experimental data points. All public data within ECBD can be accessed through its user interface, API or by database dump under the CC-BY 4.0 license.

BindingDB in 2024: a FAIR knowledgebase of protein-small molecule binding data

BindingDB (bindingdb.org) is a public, web-accessible database of experimentally measured binding affinities between small molecules and proteins, which supports diverse applications including medicinal chemistry, biochemical pathway annotation, training of artificial intelligence models and computational chemistry methods development. This update reports significant growth and enhancements since our last review in 2016. Of note, the database now contains 2.9 million binding measurements spanning 1.3 million compounds and thousands of protein targets. This growth is largely attributable to our unique focus on curating data from US patents, which has yielded a substantial influx of novel binding data. Recent improvements include a remake of the website following responsive web design principles, enhanced search and filtering capabilities, new data download options and webservices and establishment of a long-term data archive replicated across dispersed sites. We also discuss BindingDB's positioning relative to related resources, its open data sharing policies, insights gleaned from the dataset and plans for future growth and development.

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/.

DIONYSUS: a database of protein-carbohydrate interfaces

Protein-carbohydrate interactions govern a wide variety of biological processes and play an essential role in the development of different diseases. Here, we present DIONYSUS, the first database of protein-carbohydrate interfaces annotated according to structural, chemical and functional properties of both proteins and carbohydrates. We provide exhaustive information on the nature of interactions, binding site composition, biological function and specific additional information retrieved from existing databases. The user can easily search the database using protein sequence and structure information or by carbohydrate binding site properties. Moreover, for a given interaction site, the user can perform its comparison with a representative subset of non-covalent protein-carbohydrate interactions to retrieve information on its potential function or specificity. Therefore, DIONYSUS is a source of valuable information both for a deeper understanding of general protein-carbohydrate interaction patterns, for annotation of the previously unannotated proteins and for such applications as carbohydrate-based drug design. DIONYSUS is freely available at www.dsimb.inserm.fr/DIONYSUS/.

Chem(Pro)2: the atlas of chemoproteomic probes labelling human proteins

Chemoproteomic probes (CPPs) have been widely considered as powerful molecular biological tools that enable the highly efficient discovery of both binding proteins and modes of action for the studied compounds. They have been successfully used to validate targets and identify binders. The design of CPP has been considered extremely challenging, which asks for the generalization using a large number of probe data. However, none of the existing databases gives such valuable data of CPPs. Herein, a database entitled 'Chem(Pro)2' was therefore developed to systematically describe the atlas of diverse types of CPPs labelling human protein in living cell/lysate. With the booming application of chemoproteomic technique and artificial intelligence in current chemical biology study, Chem(Pro)2 was expected to facilitate the AI-based learning of interacting pattern among molecules for discovering innovative targets and new drugs. Till now, Chem(Pro)2 has been open to all users without any login requirement at: https://idrblab.org/chemprosquare/.

Development of a machine learning-based target-specific scoring function for structure-based binding affinity prediction for human dihydroorotate dehydrogenase inhibitors

Human dihydroorotate dehydrogenase (hDHODH) is a flavin mononucleotide-dependent enzyme that can limit de novo pyrimidine synthesis, making it a therapeutic target for diseases such as autoimmune disorders and cancer. In this study, using the docking structures of complexes generated by AutoDock Vina, we integrate interaction features and ligand features, and employ support vector regression to develop a target-specific scoring function for hDHODH (TSSF-hDHODH). The Pearson correlation coefficient values of TSSF-hDHODH in the cross-validation and external validation are 0.86 and 0.74, respectively, both of which are far superior to those of classic scoring function AutoDock Vina and random forest (RF) based generic scoring function RF-Score. TSSF-hDHODH is further used for the virtual screening of potential inhibitors in the FDA-Approved & Pharmacopeia Drug Library. In conjunction with the results from molecular dynamics simulations, crizotinib is identified as a candidate for subsequent structural optimization. This study can be useful for the discovery of hDHODH inhibitors and the development of scoring functions for additional targets.

Discovering natural products as potential inhibitors of SARS-CoV-2 spike proteins

The ongoing global pandemic caused by the SARS-CoV-2 virus has demanded the urgent search for effective therapeutic interventions. In response, our research aimed at identifying natural products (NPs) with potential inhibitory effects on the entry of the SARS-CoV-2 spike (S) protein into host cells. Utilizing the Protein Data Bank Japan (PDBJ) and BindingDB databases, we isolated 204 S-glycoprotein sequences and conducted a clustering analysis to identify similarities and differences among them. We subsequently identified 33,722 binding molecules (BMs) by matching them with the sequences of 204 S-glycoproteins and compared them with 52,107 secondary metabolites (SMs) from the KNApSAcK database to identify potential inhibitors. We conducted docking and drug-likeness property analyses to identify several SMs with potential as drug candidates based on binding energy (BE), no Lipinski's rule violation (LV), psychochemical properties within the pink area of the bioavailability radar, and a bioavailability score (BAS) not less than 0.55. Fourteen SMs were predicted through computational analysis as potential candidates for inhibiting the three major types of S proteins. Our study provides a foundation for further experimental validation of these compounds as potential therapeutic agents against SARS-CoV-2.

Network Medicine-Based Strategy Identifies Maprotiline as a Repurposable Drug by Inhibiting PD-L1 Expression via Targeting SPOP in Cancer

Immune checkpoint inhibitors (ICIs) are drugs that inhibit immune checkpoint (ICP) molecules to restore the antitumor activity of immune cells and eliminate tumor cells. Due to the limitations and certain side effects of current ICIs, such as programmed death protein-1, programmed cell death-ligand 1, and cytotoxic T lymphocyte-associated antigen 4 (CTLA4) antibodies, there is an urgent need to find new drugs with ICP inhibitory effects. In this study, a network-based computational framework called multi-network algorithm-driven drug repositioning targeting ICP (Mnet-DRI) is developed to accurately repurpose novel ICIs from ≈3000 Food and Drug Administration-approved or investigational drugs. By applying Mnet-DRI to PD-L1, maprotiline (MAP), an antidepressant drug is repurposed, as a potential PD-L1 modifier for colorectal and lung cancers. Experimental validation revealed that MAP reduced PD-L1 expression by targeting E3 ubiquitin ligase speckle-type zinc finger structural protein (SPOP), and the combination of MAP and anti-CTLA4 in vivo significantly enhanced the antitumor effect, providing a new alternative for the clinical treatment of colorectal and lung cancer.

hERGBoost: A gradient boosting model for quantitative IC50 prediction of hERG channel blockers

The human ether-a-go-go-related gene (hERG) potassium channel is pivotal in drug discovery due to its susceptibility to blockage by drug candidate molecules, which can cause severe cardiotoxic effects. Consequently, identifying and excluding potential hERG channel blockers at the earliest stages of drug development is crucial. Most traditional machine learning models predict a molecule's cardiotoxicity or non-cardiotoxicity typically at 10 μM, which doesn't account for compounds with low IC50 values that are non-toxic at therapeutic levels due to their high effectiveness at lower concentrations. To address the need for more precise, quantitative predictions, we developed hERGBoost, a cutting-edge machine learning model employing a gradient-boosting algorithm. This model demonstrates superior accuracy in predicting the IC50 of drug candidates. Trained on a specially curated dataset for this study, hERGBoost not only exhibited excellent performance in external validation, achieving an R2 score of 0.394 and a low root mean square error of 0.616 but also significantly outstripped previous models in both qualitative and quantitative assessments. Representing a notable leap forward in the prediction of hERG channel blockers, the hERGBoost model and its datasets are freely available to the drug discovery community on our web server at. http://ssbio.cau.ac.kr/software/hergboost This resource promises to be invaluable in advancing safer pharmaceutical development.

Leveraging Artificial Intelligence in GPCR Activation Studies: Computational Prediction Methods as Key Drivers of Knowledge

G protein-coupled receptors (GPCRs) are key molecules involved in cellular signaling and are attractive targets for pharmacological intervention. This chapter is designed to explore the range of algorithms used to predict GPCRs' activation states, while also examining the pharmaceutical implications of these predictions. Our primary objective is to show how artificial intelligence (AI) is key in GPCR research to reveal the intricate dynamics of activation and inactivation processes, shedding light on the complex regulatory mechanisms of this vital protein family. We describe several computational strategies that leverage diverse structural data from the Protein Data Bank, molecular dynamic simulations, or ligand-based methods to predict the activation states of GPCRs. We demonstrate how the integration of AI into GPCR research not only enhances our understanding of their dynamic properties but also presents immense potential for driving pharmaceutical research and development, offering promising new avenues in the search for newer, better therapeutic agents.

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.

Estimating AChE inhibitors from MCE database by machine learning and atomistic calculations

Acetylcholinesterase (AChE) is one of the most successful targets for the treatment of Alzheimer's disease (AD). Inhibition of AChE can result in preventing AD. In this context, the machine-learning (ML) model, molecular docking, and molecular dynamics calculations were employed to characterize the potential inhibitors for AChE from MedChemExpress (MCE) database. The trained ML model was initially employed for estimating the inhibitory of MCE compounds. Atomistic simulations including molecular docking and molecular dynamics simulations were then used to confirm ML outcomes. In particular, the physical insights into the ligand binding to AChE were clarified over the calculations. Two compounds, PubChem ID of 130467298 and 132020434, were indicated that they can inhibit AChE.

Exploring natural products potential: A similarity-based target prediction tool for natural products

Natural products are invaluable resources in drug discovery due to their substantial structural diversity. However, predicting their interactions with druggable protein targets remains a challenge, primarily due to the limited availability of bioactivity data. This study introduces CTAPred (Compound-Target Activity Prediction), an open-source command-line tool designed to predict potential protein targets for natural products. CTAPred employs a two-stage approach, combining fingerprinting and similarity-based search techniques to identify likely drug targets for these bioactive compounds. Despite its simplicity, the tool's performance is comparable to that of more complex methods, demonstrating proficiency in target retrieval for natural product compounds. Furthermore, this study explores the optimal number of reference compounds most similar to the query compound, aiming to refine target prediction accuracy. The findings demonstrated the superior performance of considering only the most similar reference compounds for target prediction. CTAPred is freely available at https://github.com/Alhasbary/CTAPred, offering a valuable resource for deciphering natural product-target associations and advancing drug discovery.

ChemNTP: Advanced Prediction of Neurotoxicity Targets for Environmental Chemicals Using a Siamese Neural Network

Environmental chemicals can enter the human body through various exposure pathways, potentially leading to neurotoxic effects that pose significant health risks. Many such chemicals have been identified as neurotoxic, but the molecular mechanisms underlying their toxicity, including specific binding targets, remain unclear. To address this, we developed ChemNTP, a predictive model for identifying neurotoxicity targets of environmental chemicals. ChemNTP integrates a comprehensive representation of chemical structures and biological targets, improving upon traditional methods that are limited to single targets and mechanisms. By leveraging these structural representations, ChemNTP enables rapid screening across 199 potential neurotoxic targets or key molecular initiating events (MIEs). The model demonstrates robust predictive performance, achieving an area under the receiver operating characteristic curve (AUCROC) of 0.923 on the test set. Additionally, ChemNTP's attention mechanism highlights critical residues in binding targets and key functional groups or atoms in molecules, offering insights into the structural basis of interactions. Experimental validation through in vitro enzyme activity assays and molecular docking confirmed the binding of eight polybrominated diphenyl ethers (PBDEs) to acetylcholinesterase (AChE). We also provide a user-friendly software interface to facilitate the rapid identification of neurotoxicity targets for emerging environmental pollutants, with potential applications in studying MIEs for more types of toxicity.

e-QSAR (Explainable AI-QSAR), molecular docking, and ADMET analysis of structurally diverse GSK3-beta modulators to identify concealed modulatory features vindicated by X-ray

Glycogen Synthase Kinase-3 beta (GSK-3β) is a crucial enzyme linked to various cellular processes, including neurodegeneration, autophagy, and diabetes. A structurally diverse set of 1293 molecules having GSK-3β modulatory activity has been used. Molecular docking and eXplainable Artificial Intelligence (XAI) have been used concomitantly. The approach involves using GA for feature selection and XGBoost for in-depth analysis, yielding strong statistical validation with R2tr = 0.9075, R2L10 %O = 0.9116, and Q2F3 = 0.7841. Molecular docking provided complementary and similar results. Machine learning model interpretation using SHapley Additive exPlanations (SHAP) revealed that specific structural features like aromatic carbon with specific partial charges, non-ring nitrogen atoms, sp3-hybrid carbon atoms, and the topological distance between carbon and nitrogen atoms, among others, significantly influence the modulatory profile. The results are also supported by reported X-ray resolved structures. In addition, in-silico ADMET analysis is also accomplished. This research underscores the value of advanced machine learning techniques in understanding complex biological phenomena and supporting rational drug design.

From SMILES to Enhanced Molecular Property Prediction: A Unified Multimodal Framework with Predicted 3D Conformers and Contrastive Learning Techniques

We present a novel molecular property prediction framework that requires only the SMILES format as input but is designed to be multimodal by incorporating predicted 3D conformer representations. Our model captures comprehensive molecular features by leveraging both the sequential character structure of SMILES and the three-dimensional spatial structure of conformers. The framework employs contrastive learning techniques, utilizing InfoNCE loss to align SMILES and conformer embeddings, along with task-specific loss functions, such as ConR for regression and SupCon for classification. To address data imbalance, we incorporate feature distribution smoothing (FDS), a common challenge in drug discovery. We evaluated the framework through multiple case studies, including SARS-CoV-2 drug docking score prediction, molecular property prediction using MoleculeNet data sets, and kinase inhibitor prediction for JAK-1, JAK-2, and MAPK-14 using custom data sets curated from PubChem. The results consistently outperformed state-of-the-art methods, with ConR and FDS significantly improving regression tasks and SupCon enhancing classification performance. These findings highlight the flexibility and robustness of our multimodal model, demonstrating its effectiveness across diverse molecular property prediction tasks, with promising applications in drug discovery and molecular analysis.

AttenhERG: a reliable and interpretable graph neural network framework for predicting hERG channel blockers

Cardiotoxicity, particularly drug-induced arrhythmias, poses a significant challenge in drug development, highlighting the importance of early-stage prediction of human ether-a-go-go-related gene (hERG) toxicity. hERG encodes the pore-forming subunit of the cardiac potassium channel. Traditional methods are both costly and time-intensive, necessitating the development of computational approaches. In this study, we introduce AttenhERG, a novel graph neural network framework designed to predict hERG channel blockers reliably and interpretably. AttenhERG demonstrates improved performance compared to existing methods with an AUROC of 0.835, showcasing its efficacy in accurately predicting hERG activity across diverse datasets. Additionally, uncertainty evaluation analysis reveals the model's reliability, enhancing its utility in drug discovery and safety assessment. Case studies illustrate the practical application of AttenhERG in optimizing compounds for hERG toxicity, highlighting its potential in rational drug design.Scientific contributionAttenhERG is a breakthrough framework that significantly improves the interpretability and accuracy of predicting hERG channel blockers. By integrating uncertainty estimation, AttenhERG demonstrates superior reliability compared to benchmark models. Two case studies, involving APH1A and NMT1 inhibitors, further emphasize AttenhERG's practical application in compound optimization.

Variations in metabolite fingerprints of Tinospora species targeting metabolic disorders: an integrated metabolomics and network pharmacology approach

INTRODUCTION

Metabolic disorders are a global health concern, necessitating the development of drugs with fewer side effects and more efficacy. Traditional Indian medicine uses Tinospora cordifolia and Tinospora sinensis, but their metabolite fingerprints and impact on geographical location remains unknown.

OBJECTIVE

The present study aimed to identify metabolite fingerprints from T. cordifolia and T. sinensis species from different geographic locations and also to identify potential quality markers for treating metabolic disorders.

METHODS

Non-targeted metabolite fingerprinting of T. cordifolia and T. sinensis was performed using HPLC-QTOF-MS/MS analysis. Network pharmacology, molecular docking and molecular dynamics simulation analysis were performed to identify potential quality markers, hub targets, and key pathways associated with metabolic disorders.

RESULTS

In this study, six potential marker compounds and twenty-five differential compounds were identified between T. cordifolia and T. sinensis. Based on geography, five and one metabolite marker compounds were identified in T. cordifolia and T. sinensis respectively. Network pharmacology, molecular docking, and molecular dynamics simulation analysis revealed trans piceid, crustecdysone in T. cordifolia, and gallic acid in T. sinensis as potential quality markers against metabolic disorder related hub targets.

CONCLUSION

Integration of non-targeted metabolomics and network pharmacology approach deciphers the pharmacological mechanism of action in terms of identifying potential quality markers from Tinospora species that can be used against metabolic disorders. However, further research is required to validate these findings in in vitro and in vivo studies for better assertion.

Hiding in plain sight: Optimizing topoisomerase IIα inhibitors into Hsp90β selective binders

Due to their impact on several oncogenic client proteins, the Hsp90 family of chaperones has been widely studied for the development of potential anticancer agents. Although several Hsp90 inhibitors have entered clinical trials, most were unsuccessful because they induced a heat shock response (HSR). This issue can be circumvented by using isoform-selective inhibitors, but the high similarity in the ATP-binding sites between the isoforms presents a challenge. Given that Hsp90 shares a conserved Bergerat fold with bacterial DNA gyrase B and human topoisomerase IIα, we repurposed our ATP-competitive inhibitors of these two proteins for Hsp90 inhibition. We virtually screened a library of in-house inhibitors and identified eleven hits for evaluation of Hsp90 binding. Among these, compound 11 displayed low micromolar affinity for Hsp90 and demonstrated a 12-fold selectivity for Hsp90β over its closest isoform, Hsp90α. Out of 29 prepared analogs, 16 showed a preference for Hsp90β over Hsp90α. Furthermore, eleven of these compounds inhibited the growth of several cancer cell lines in vitro. Notably, compound 24e reduced intracellular levels of Hsp90 client proteins in MCF-7 cells, leading to cell cycle arrest in the G0/G1 phase without inducing HSR. This inhibitor exhibited at least a 27-fold preference for Hsp90β and was selective against topoisomerase IIα, a panel of 22 representative protein kinases, and proved to be non-toxic in a zebrafish larvae toxicology model. Finally, molecular modeling, corroborated by STD NMR studies, and the binding of 24e to the S52A mutant of Hsp90α confirmed that the serine to alanine switch drives the selectivity between the two cytoplasmic isoforms.

Improved Prediction of Ligand-Protein Binding Affinities by Meta-modeling

The accurate screening of candidate drug ligands against target proteins through computational approaches is of prime interest to drug development efforts. Such virtual screening depends in part on methods to predict the binding affinity between ligands and proteins. Many computational models for binding affinity prediction have been developed, but with varying results across targets. Given that ensembling or meta-modeling approaches have shown great promise in reducing model-specific biases, we develop a framework to integrate published force-field-based empirical docking and sequence-based deep learning models. In building this framework, we evaluate many combinations of individual base models, training databases, and several meta-modeling approaches. We show that many of our meta-models significantly improve affinity predictions over base models. Our best meta-models achieve comparable performance to state-of-the-art deep learning tools exclusively based on 3D structures while allowing for improved database scalability and flexibility through the explicit inclusion of features such as physicochemical properties or molecular descriptors. We further demonstrate improved generalization capability by our models using a large-scale benchmark of affinity prediction as well as a virtual screening application benchmark. Overall, we demonstrate that diverse modeling approaches can be ensembled together to gain meaningful improvement in binding affinity prediction.

CSearch: chemical space search via virtual synthesis and global optimization

The two key components of computational molecular design are virtually generating molecules and predicting the properties of these generated molecules. This study focuses on an effective method for molecular generation through virtual synthesis and global optimization of a given objective function. Using a pre-trained graph neural network (GNN) objective function to approximate the docking energies of compounds for four target receptors, we generated highly optimized compounds with 300-400 times less computational effort compared to virtual compound library screening. These optimized compounds exhibit similar synthesizability and diversity to known binders with high potency and are notably novel compared to library chemicals or known ligands. This method, called CSearch, can be effectively utilized to generate chemicals optimized for a given objective function. With the GNN function approximating docking energies, CSearch generated molecules with predicted binding poses to the target receptors similar to known inhibitors, demonstrating its effectiveness in producing drug-like binders.Scientific Contribution We have developed a method for effectively exploring the chemical space of drug-like molecules using a global optimization algorithm with fragment-based virtual synthesis. The compounds generated using this method optimize the given objective function efficiently and are synthesizable like commercial library compounds. Furthermore, they are diverse, novel drug-like molecules with properties similar to known inhibitors for target receptors.

Combining crystallographic and binding affinity data towards a novel dataset of small molecule overlays

Although small molecule superposition is a standard technique in drug discovery, a rigorous performance assessment of the corresponding methods is currently challenging. Datasets in this field are sparse, small, tailored to specific applications, unavailable, or outdated. The newly developed LOBSTER set described herein offers a publicly available and method-independent dataset for benchmarking and method optimization. LOBSTER stands for "Ligand Overlays from Binding SiTe Ensemble Representatives". All ligands were derived from the PDB in a fully automated workflow, including a ligand efficiency filter. So-called ligand ensembles were assembled by aligning identical binding sites. Thus, the ligands within the ensembles are superimposed according to their experimentally determined binding orientation and conformation. Overall, 671 representative ligand ensembles comprise 3583 ligands from 3521 proteins. Altogether, 72,734 ligand pairs based on the ensembles were grouped into ten distinct subsets based on their volume overlap, for the benefit of introducing different degrees of difficulty for evaluating superposition methods. Statistics on the physicochemical properties of the compounds indicate that the dataset represents drug-like compounds. Consensus Diversity Plots show predominantly high Bemis-Murcko scaffold diversity and low median MACCS fingerprint similarity for each ensemble. An analysis of the underlying protein classes further demonstrates the heterogeneity within our dataset. The LOBSTER set offers a variety of applications like benchmarking multiple as well as pairwise alignments, generating training and test sets, for example based on time splits, or empirical software performance evaluation studies. The LOBSTER set is publicly available at https://doi.org/10.5281/zenodo.12658320 , representing a stable and versioned data resource. The Python scripts are available at https://github.com/rareylab/LOBSTER , open-source, and allow for updating or recreating superposition sets with different data sources.

Comparative Metabolomics and Network Pharmacology Analysis Reveal Shared Neuroprotective Mechanisms of Bacopa monnieri (L.) Wettst and Centella asiatica (L.) Urb

Bacopa monnieri (L.) Wettst and Centella asiatica (L.) Urb., two nootropics, are recognized in Indian Ayurvedic texts. Studies have attempted to understand their action as memory enhancers and neuroprotectants, but many molecular aspects remain unknown. We propose that Bacopa monnieri (L.) Wettst and Centella asiatica (L.) Urb. share common neuroprotective mechanisms. Mass spectrometry-based untargeted metabolomics and network pharmacology approach were used to identify potential protein targets for the metabolites from each extract. Phytochemical analyses and cell culture validation studies were also used to assess apoptosis and ROS activity using aqueous extracts prepared from both herbal powders. Further, docking studies were also performed using the LibDock protocol. Untargeted metabolomics and network pharmacology approach unveiled 2751 shared metabolites and 3439 and 2928 non-redundant metabolites from Bacopa monnieri and Centella asiatica extracts, respectively, suggesting a potential common neuroprotective mechanism among these extracts. Protein-target prediction highlighted 92.4% similarity among the proteins interacting with metabolites for these extracts. Among them, kinases mapped to MAPK, mTOR, and PI3K-AKT signaling pathways represented a predominant population. Our results highlight a significant similarity in the metabolome of Bacopa monnieri (L.) Wettst and Centella asiatica (L.) Urb., and their potential protein targets may be attributed to their common neuroprotective functions.

Screening of BindingDB database ligands against EGFR, HER2, Estrogen, Progesterone and NF-κB receptors based on machine learning and molecular docking

Breast cancer, the second most prevalent cancer among women worldwide, necessitates the exploration of novel therapeutic approaches. To target the four subgroups of breast cancer "hormone receptor-positive and HER2-negative, hormone receptor-positive and HER2-positive, hormone receptor-negative and HER2-positive, and hormone receptor-negative and HER2-negative" it is crucial to inhibit specific targets such as EGFR, HER2, ER, NF-κB, and PR. In this study, we evaluated various methods for binary and multiclass classification. Among them, the GA-SVM-SVM:GA-SVM-SVM model was selected with an accuracy of 0.74, an F1-score of 0.73, and an AUC of 0.92 for virtual screening of ligands from the BindingDB database. This model successfully identified 4454, 803, 438, and 378 ligands with over 90% precision in both active/inactive and target prediction for the classes of EGFR+HER2, ER, NF-κB, and PR, respectively, from the BindingDB database. Based on to the selected ligands, we created a dendrogram that categorizes different ligands based on their targets. This dendrogram aims to facilitate the exploration of chemical space for various therapeutic targets. Ligands that surpassed a 90% threshold in the product of activity probability and correct target selection probability were chosen for further investigation using molecular docking. The binding energy range for these ligands against their respective targets was calculated to be between -15 and -5 kcal/mol. Finally, based on general and common rules in medicinal chemistry, we selected 2, 3, 3, and 8 new ligands with high priority for further studies in the EGFR+HER2, ER, NF-κB, and PR classes, respectively.

Data-augmented machine learning scoring functions for virtual screening of YTHDF1 m6A reader protein

Machine learning is rapidly advancing the drug discovery process, significantly enhancing speed and efficiency. Innovation in computer-aided drug design is primarily driven by structure- and ligand-based approaches. When the number of known inhibitors for a target is limited, data augmentation strategies are often preferred to enhance model performance. In this study, we developed predictive machine learning models for structure-based drug discovery leveraging multiple traditional machine learning algorithms trained with target and ligand dynamics-aware datasets. To illustrate our approach, we present a composite model that combines classification and regression to predict YTHDF1 inhibitors, utilizing PLEC features. YTHDF1, a key m6A reader protein involved in mRNA translation, is implicated in various cancers, making it a promising therapeutic target. Traditional structure-based virtual screening (SBVS) using generic scoring functions has struggled to identify potent YTHDF1 inhibitors due to the protein's unique binding characteristics. To overcome this, we developed YTHDF1-specific machine learning scoring functions (MLSFs) to enhance SBVS efficacy. We employed various data augmentation techniques to generate a comprehensive dataset, incorporating multiple conformations of ligands and the YTHDF1 protein. We have trained 64 YTHDF1-specific MLSFs using four machine learning algorithms and evaluated them on ten test sets, focusing on their predictive and ranking power. Our results demonstrate that the artificial neural network with protein-ligand extended connectivity fingerprints (ANN-PLEC) outperforms other MLSFs, consistently achieving high area under the precision-recall curve (PR-AUC) of 0.87. This method shows promise for targets with limited quantities of active molecules, providing a viable path forward for drug discovery research. The ANN-PLEC scoring function is made freely available on GitHub for other researchers to access and utilize https://github.com/JuniML/SBVS-YTHDF1/.

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.

Network-based identification and mechanism exploration of active ingredients against Alzheimer's disease via targeting endoplasmic reticulum stress from traditional chinese medicine

Alzheimer's disease is a neurodegenerative disease that leads to dementia and poses a serious threat to the health of the elderly. Traditional Chinese medicine (TCM) presents as a promising novel therapeutic therapy for preventing and treating dementia. Studies have shown that natural products derived from kidney-tonifying herbs can effectively inhibit AD. Furthermore, endoplasmic reticulum (ER) stress is a critical factor in the pathology of AD. Regulation of ER stress is a crucial approach to prevent and treat AD. Thus, in this study, we first collected kidney-tonifying herbs, integrated chemical ingredients from multiple TCM databases, and constructed a comprehensive drug-target network. Subsequently, we employed the endophenotype network (network proximity) method to identify potential active ingredients in kidney-tonifying herbs that prevented AD via regulating ER stress. By combining the predicted outcomes, we discovered that 32 natural products could ameliorate AD pathology via regulating ER stress. After a comprehensive evaluation of the multi-network model and systematic pharmacological analyses, we further selected several promising compounds for in vitro testing in the APP-SH-SY5Y cell model. Experimental results showed that echinacoside and danthron were able to effectively reduce ER stress-mediated neuronal apoptosis by inhibiting the expression levels of BIP, p-PERK, ATF6, and CHOP in APP-SH-SY5Y cells. Overall, this study utilized the endophenotype network to preliminarily decipher the effective material basis and potential molecular mechanism of kidney-tonifying Chinese medicine for prevention and treatment of AD.

NaCTR: Natural product-derived compound-based drug discovery pipeline from traditional oriental medicine by search space reduction

The drug discovery pipelines require enormous time and cost, albeit their infamously high risk of failures. Reducing such risk has therefore been the utmost goal in the process. Recently, natural products (NPs) in traditional oriental medicine (TOM) have come into the spotlight for their efficacy and safety supported throughout the history. Not only that, with the ever-increasing repository of various biological datasets, many data-driven in silico approaches have also been extensively studied for better efficient search and testing. However, TOM-based datasets lack information on recently prevalent diseases, while experimental datasets are prone to provide target spaces that are too large. Adequate combination of both approaches can therefore fill in each other's blanks. In this study, we introduce NaCTR, an in silico discovery pipeline that achieves such integration to suggest NPs-derived drug candidates for a given disease. First, phenotypes and disease genes for the disease are identified in literature and public databases. Secondly, a pool of potentially therapeutic NPs are identified based on both TOM-based phenotype records and compound-gene interaction datasets. Lastly, the compounds contained in the NPs are further screened for toxicity and pharmacokinetic properties. We use the Parkinson's disease as the case study to test the NaCTR pipeline. Through the pipeline, we propose glutathione and four other compounds as novel drug candidates. We further highlight the finding with literature support. As the first to effectively combine data from ancient and recent repositories, the NaCTR pipeline can be a novel pipeline that can be applied successfully to any other diseases.

A review of machine learning-based methods for predicting drug-target interactions

The prediction of drug-target interactions (DTI) is a crucial preliminary stage in drug discovery and development, given the substantial risk of failure and the prolonged validation period associated with in vitro and in vivo experiments. In the contemporary landscape, various machine learning-based methods have emerged as indispensable tools for DTI prediction. This paper begins by placing emphasis on the data representation employed by these methods, delineating five representations for drugs and four for proteins. The methods are then categorized into traditional machine learning-based approaches and deep learning-based ones, with a discussion of representative approaches in each category and the introduction of a novel taxonomy for deep neural network models in DTI prediction. Additionally, we present a synthesis of commonly used datasets and evaluation metrics to facilitate practical implementation. In conclusion, we address current challenges and outline potential future directions in this research field.

CASTER-DTA: Equivariant Graph Neural Networks for Predicting Drug-Target Affinity

Accurately determining the binding affinity of a ligand with a protein is important for drug design, development, and screening. With the advent of accessible protein structure prediction methods such as AlphaFold, several approaches have been developed that make use of information determined from the 3D structure for a variety of downstream tasks. However, methods for predicting binding affinity that do consider protein structure generally do not take full advantage of such 3D structural protein information, often using such information only to define nearest-neighbor graphs based on inter-residue or inter-atomic distances. Here, we present a joint architecture that we call CASTER-DTA (Cross-Attention with Structural Target Equivariant Representations for Drug-Target Affinity) that makes use of an SE(3)-equivariant graph neural network to learn more robust protein representations alongside a standard graph neural network to learn molecular representations, and we further augment these representations by incorporating an attention-based mechanism by which individual residues in a protein can attend to atoms in a ligand and vice-versa to improve interpretability. In this manner, we show that using equivariant graph neural networks in our architecture enables CASTER-DTA to approach and exceed state-of-the-art performance in predicting drug-target affinity without the inclusion of external information, such as protein language model embeddings. We do so on the Davis and KIBA datasets, common benchmarks for predicting drug-target affinity. We also discuss future steps to further improve performance.

Accurate and transferable drug-target interaction prediction with DrugLAMP

MOTIVATION

Accurate prediction of drug-target interactions (DTIs), especially for novel targets or drugs, is crucial for accelerating drug discovery. Recent advances in pretrained language models (PLMs) and multi-modal learning present new opportunities to enhance DTI prediction by leveraging vast unlabeled molecular data and integrating complementary information from multiple modalities.

RESULTS

We introduce DrugLAMP (PLM-assisted multi-modal prediction), a PLM-based multi-modal framework for accurate and transferable DTI prediction. DrugLAMP integrates molecular graph and protein sequence features extracted by PLMs and traditional feature extractors. We introduce two novel multi-modal fusion modules: (i) pocket-guided co-attention (PGCA), which uses protein pocket information to guide the attention mechanism on drug features, and (ii) paired multi-modal attention (PMMA), which enables effective cross-modal interactions between drug and protein features. These modules work together to enhance the model's ability to capture complex drug-protein interactions. Moreover, the contrastive compound-protein pre-training (2C2P) module enhances the model's generalization to real-world scenarios by aligning features across modalities and conditions. Comprehensive experiments demonstrate DrugLAMP's state-of-the-art performance on both standard benchmarks and challenging settings simulating real-world drug discovery, where test drugs/targets are unseen during training. Visualizations of attention maps and application to predict cryptic pockets and drug side effects further showcase DrugLAMP's strong interpretability and generalizability. Ablation studies confirm the contributions of the proposed modules.

AVAILABILITY AND IMPLEMENTATION

Source code and datasets are freely available at https://github.com/Lzcstan/DrugLAMP. All data originate from public sources.

In silico screening of phytochemicals against chromatin modifier, SETD7 for remodeling of the immunosuppressive tumor microenvironment in renal cancer

The tumor microenvironment and immune evasion function in a complex cellular network profoundly challenge the clinical outcome of promising therapies. Our recently published study reported that the subset of genes upregulated in ccRCC due to H3K4me1 and DNA demethylation potentially leads to an immunosuppressive environment. Thus, modulating H3K4me1 chromatin modifier SETD7 with a natural inhibitor in combination with immunotherapy might improve the immune landscape for a better therapeutic outcome. The present study was conducted via virtual screening and MD simulation using compounds from the literature, IMPPAT, and SuperNatural database. The phytochemical IMPHY002979 showed better binding affinity and lower energy than the reported R-PFI-2 and cyproheptadine inhibitors. The phytochemicals interact with the SET domain through H-bonding, as confirmed by MD simulation and molecular interaction analysis. Further, the compound was assessed using ADME parameters and free energy estimation, showing better pharmacokinetic properties. Therefore, the non-accessibility of the histone methyltransferase activity domain of SET7 with IMPHY002979 can downregulate H3K4me1 and, thereby, the expression of genes potentially responsible for immunosuppressive TME. Thus, patient stratification based on molecular markers for immunotherapy and combining epigenetic modulators with therapeutic drugs will improve the efficacy of immunotherapy in ccRCC.

Mixture-of-Experts Based Dissociation Kinetic Model for De Novo Design of HSP90 Inhibitors with Prolonged Residence Time

The dissociation rate constant (koff) significantly impacts the drug potency and dosing frequency. This work proposes a powerful optimization-based framework for de novo drug design guided by koff. First, a comprehensive database containing 2,773 unique koff values is created. Based on the database, a novel generic dissociation kinetic model is developed with a mixture-of-experts architecture, enabling high-throughput predictions of koff with high accuracy. The developed model is then integrated with an optimization-based mathematical programming approach to design drug candidates with low koff. Finally, the τ-RAMD method is utilized to rigorously verify the designed potential drug candidates. In a case study, the framework successfully identified numerous new potential HSP90 inhibitor candidates, achieving a maximum 45.7% improvement in residence time (τ = 1/koff) compared to that of a known exceptional HSP90 inhibitor. These findings demonstrate the feasibility and effectiveness of the kinetics-guided optimization-based de novo drug design framework in designing drug candidates with prolonged τ.

Ultra-high-throughput screening of antimicrobial combination therapies using a two-stage transparent machine learning model

Here, we present M2D2, a two-stage machine learning (ML) pipeline that identifies promising antimicrobial drug combinations, which are crucial for combating drug resistance. M2D2 addresses key challenges in drug combination discovery by predicting drug synergies using computationally generated drug-protein interaction data, thereby circumventing the need for expensive omics data. The model improves the accuracy of drug target identification using high-throughput experimental and computational methods via feedback between ML stages. M2D2's transparent framework provides mechanistic insights into drug interactions and was benchmarked against chemogenomics, transcriptomics, and metabolomics datasets. We experimentally validated M2D2 using high-throughput screening of 946 combinations of Food and Drug Administration (FDA)- approved drugs and antibiotics against Escherichia coli . We discovered synergy between a cerebrovascular drug and a widely used penicillin antibiotic and validated predicted mechanisms of action using genome-wide CRISPR inhibition screens. M2D2 offers a transparent ML tool for rapidly designing combination therapies and guides repurposing efforts while providing mechanistic insights.

ClickGen: Directed exploration of synthesizable chemical space via modular reactions and reinforcement learning

Despite the significant potential of generative models, low synthesizability of many generated molecules limits their real-world applications. In response to this issue, we develop ClickGen, a deep learning model that utilizes modular reactions like click chemistry to assemble molecules and incorporates reinforcement learning along with inpainting technique to ensure that the proposed molecules display high diversity, novelty and strong binding tendency. ClickGen demonstrates superior performance over the other reaction-based generative models in terms of novelty, synthesizability, and docking conformation similarity for existing binders targeting the three proteins. We then proceeded to conduct wet-lab validation on the ClickGen's proposed molecules for poly adenosine diphosphate-ribose polymerase 1. Due to the guaranteed high synthesizability and model-generated synthetic routes for reference, we successfully produced and tested the bioactivity of these novel compounds in just 20 days, much faster than typically expected time frame when handling sufficiently novel molecules. In bioactivity assays, two lead compounds demonstrated superior anti-proliferative efficacy against cancer cell lines, low toxicity, and nanomolar-level inhibitory activity to PARP1. We demonstrate that ClickGen and related models may represent a new paradigm in molecular generation, bringing AI-driven, automated experimentation and closed-loop molecular design closer to realization.

Drug-target interaction prediction by integrating heterogeneous information with mutual attention network

BACKGROUND

Identification of drug-target interactions is an indispensable part of drug discovery. While conventional shallow machine learning and recent deep learning methods based on chemogenomic properties of drugs and target proteins have pushed this prediction performance improvement to a new level, these methods are still difficult to adapt to novel structures. Alternatively, large-scale biological and pharmacological data provide new ways to accelerate drug-target interaction prediction.

METHODS

Here, we propose DrugMAN, a deep learning model for predicting drug-target interaction by integrating multiplex heterogeneous functional networks with a mutual attention network (MAN). DrugMAN uses a graph attention network-based integration algorithm to learn network-specific low-dimensional features for drugs and target proteins by integrating four drug networks and seven gene/protein networks collected by a certain screening conditions, respectively. DrugMAN then captures interaction information between drug and target representations by a mutual attention network to improve drug-target prediction.

RESULTS

DrugMAN achieved the best performance compared with cheminformation-based methods SVM, RF, DeepPurpose and network-based deep learing methods DTINet and NeoDT in four different scenarios, especially in real-world scenarios. Compared with SVM, RF, deepurpose, DTINet, and NeoDT, DrugMAN showed the smallest decrease in AUROC, AUPRC, and F1-Score from warm-start to Both-cold scenarios. This result is attributed to DrugMAN's learning from heterogeneous data and indicates that DrugMAN has a good generalization ability. Taking together, DrugMAN spotlights heterogeneous information to mine drug-target interactions and can be a powerful tool for drug discovery and drug repurposing.

Deep learning pipeline for accelerating virtual screening in drug discovery

In the race to combat ever-evolving diseases, the drug discovery process often faces the hurdles of high-cost and time-consuming procedures. To tackle these challenges and enhance the efficiency of identifying new therapeutic agents, we introduce VirtuDockDL, which is a streamlined Python-based web platform utilizing deep learning for drug discovery. This pipeline employs a Graph Neural Network to analyze and predict the effectiveness of various compounds as potential drug candidates. During the validation phase, VirtuDockDL was instrumental in identifying non-covalent inhibitors against the VP35 protein of the Marburg virus, a critical target given the virus's high fatality rate and limited treatment options. Further, in benchmarking, VirtuDockDL achieved 99% accuracy, an F1 score of 0.992, and an AUC of 0.99 on the HER2 dataset, surpassing DeepChem (89% accuracy) and AutoDock Vina (82% accuracy). Compared to RosettaVS, MzDOCK, and PyRMD, VirtuDockDL outperformed them by combining both ligand- and structure-based screening with deep learning. While RosettaVS excels in accurate docking but lacks high-throughput screening, and PyRMD focuses on ligand-based methods without AI integration, VirtuDockDL offers superior predictive accuracy and full automation for large-scale datasets, making it ideal for comprehensive drug discovery workflows. These results underscore the tool's capability to identify high-affinity inhibitors accurately across various targets, including the HER2 protein for cancer therapy, TEM-1 beta-lactamase for bacterial infections, and the CYP51 enzyme for fungal infections like Candidiasis. To sum up, VirtuDockDL combines user-friendly interface design with powerful computational capabilities to facilitate rapid, cost-effective drug discovery and development. The integration of AI in drug discovery could potentially transform the landscape of pharmaceutical research, providing faster responses to global health challenges. The VirtuDockDL is available at https://github.com/FatimaNoor74/VirtuDockDL .