Drug Repurposing Research Articles

Improving drug repositioning with negative data labeling using large language models

IntroductionDrug repositioning offers numerous advantages, such as faster development timelines, reduced costs, and lower failure rates in drug development. Supervised machine learning is commonly used to score drug candidates but is hindered by the lack of reliable negative data—drugs that fail due to inefficacy or toxicity— which is difficult to access, lowering their prediction accuracy and generalization. Positive-Unlabeled (PU) learning has been used to overcome this issue by either randomly sampling unlabeled drugs or identifying probable negatives but still suffers from misclassification or oversimplified decision boundaries.ResultsWe proposed a novel strategy using Large Language Models (GPT-4) to analyze all clinical trials on prostate cancer and systematically identify true negatives. This approach showed remarkable improvement in predictive accuracy on independent test sets with a Matthews Correlation Coefficient of 0.76 (± 0.33) compared to 0.55 (± 0.15) and 0.48 (± 0.18) for two commonly used PU learning approaches. Using our labeling strategy, we created a training set of 26 positive and 54 experimentally validated negative drugs. We then applied a machine learning ensemble to this new dataset to assess the repurposing potential of the remaining 11,043 drugs in the DrugBank database. This analysis identified 980 potential candidates for prostate cancer. A detailed review of the top 30 revealed 9 promising drugs targeting various mechanisms such as genomic instability, p53 regulation, or TMPRSS2-ERG fusion.ConclusionBy expanding our negative data labeling approach to all diseases within the ClinicalTrials.gov database, our method could greatly advance supervised drug repositioning, offering a more accurate and data-driven path for discovering new treatments.

Advanced AI and ML frameworks for transforming drug discovery and optimization: With innovative insights in polypharmacology, drug repurposing, combination therapy and nanomedicine.

Artificial Intelligence (AI) and Machine Learning (ML) are transforming drug discovery by overcoming traditional challenges like high costs, time-consuming, and frequent failures. AI-driven approaches streamline key phases, including target identification, lead optimization, de novo drug design, and drug repurposing. Frameworks such as deep neural networks (DNNs), convolutional neural networks (CNNs), and deep reinforcement learning (DRL) models have shown promise in identifying drug targets, optimizing delivery systems, and accelerating drug repurposing. Generative adversarial networks (GANs) and variational autoencoders (VAEs) aid de novo drug design by creating novel drug-like compounds with desired properties. Case studies, such as DDR1 kinase inhibitors designed using generative models and CDK20 inhibitors developed via structure-based methods, highlight AI's ability to produce highly specific therapeutics. Models like SNF-CVAE and DeepDR further advance drug repurposing by uncovering new therapeutic applications for existing drugs. Advanced ML algorithms enhance precision in predicting drug efficacy, toxicity, and ADME-Tox properties, reducing development costs and improving drug-target interactions. AI also supports polypharmacology by optimizing multi-target drug interactions and enhances combination therapy through predictions of drug synergies and antagonisms. In nanomedicine, AI models like CURATE.AI and the Hartung algorithm optimize personalized treatments by predicting toxicological risks and real-time dosing adjustments with high accuracy. Despite its potential, challenges like data quality, model interpretability, and ethical concerns must be addressed. High-quality datasets, transparent models, and unbiased algorithms are essential for reliable AI applications. As AI continues to evolve, it is poised to revolutionize drug discovery and personalized medicine, advancing therapeutic development and patient care.

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

BackgroundThe 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.MethodsTo 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.ResultsIn 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.ConclusionsHEDDI-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.

Drug repositioning for Parkinson's disease: An emphasis on artificial intelligence approaches.

Parkinson's disease (PD) is one of the most incapacitating neurodegenerative diseases (NDDs). PD is the second most common NDD worldwide which affects approximately 1-2 percent of people over 65 years. It is an attractive pursuit for artificial intelligence (AI) to contribute to and evolve PD treatments through drug repositioning by repurposing existing drugs, shelved drugs, or even candidates that do not meet the criteria for clinical trials. A search was conducted in three databases Web of Science, Scopus, and PubMed. We reviewed the data related to the last years (1975-present) to identify those drugs currently being proposed for repositioning in PD. Moreover, we reviewed the present status of the computational approach, including AI/Machine Learning (AI/ML)-powered pharmaceutical discovery efforts and their implementation in PD treatment. It was found that the number of drug repositioning studies for PD has increased recently. Repositioning of drugs in PD is taking off, and scientific communities are increasingly interested in communicating its results and finding effective treatment alternatives for PD. A better chance of success in PD drug discovery has been made possible due to AI/ML algorithm advancements. In addition to the experimentation stage of drug discovery, it is also important to leverage AI in the planning stage of clinical trials to make them more effective. New AI-based models or solutions that increase the success rate of drug development are greatly needed.

Revolutionising Neurological Therapeutics: Investigating Drug Repurposing Strategies.

Repurposing drugs (DR) has become a viable approach to hasten the search for cures for neurodegenerative diseases (NDs). This review examines different off-target and on-target drug discovery techniques and how they might be used to find possible treatments for non-diagnostic depressions. Off-target strategies look at the known or unknown side effects of currently approved drugs for repositioning, whereas on-target strategies connect disease pathways to targets that can be treated with drugs. The review highlights the potential of experimental and computational methodologies, such as machine learning, proteomic techniques, network and genomics-based approaches, and in silico screening, in uncovering new drug-disease correlations. It also looks at difficulties and failed attempts at drug repurposing for NDs, highlighting the necessity of exact and standardised procedures to increase success rates. This review's objectives are to address the purpose of drug repurposing in human disorders, particularly neurological diseases, and to provide an overview of repurposing candidates that are presently undergoing clinical trials for neurological conditions, along with any possible causes and early findings. We then include a list of drug repurposing strategies, restrictions, and difficulties for upcoming research.

In silico Prediction of Pranlukast as a Stabilizer of PD-L1 Homodimers.

Tumors can be targeted by modulating the immune response of the patient. Programmed cell death protein 1 (PD-1) and programmed cell death ligand 1 (PD-L1) are critical immune checkpoints in cancer biology. The efficacy of certain cancer immunotherapies has been achieved by targeting these molecules using monoclonal antibodies. Small-molecule drugs have also been developed as inhibitors of the PD-1/PD-L1 axis, with a mechanism of action that is distinct from that of antibodies: they induce the formation of PD-L1 homodimers, causing their stabilization, internalization, and subsequent degradation. Drug repurposing is a strategy in which new uses are sought after for approved drugs, expediting their clinical translation based on updated findings. In this study, we generated a pharmacophore model that was based on reported small molecules that targeted PD-L1 and used it to identify potential PD-L1 inhibitors among FDA-approved drugs. We identified 12 pharmacophore-matching compounds, but only 4 reproduced the binding mode of the reference inhibitors in docking experiments. Further characterization by molecular dynamics showed that pranlukast, an antagonist of leukotriene receptors that is used to treat asthma, generated stable and energyfavorable interactions with PD-L1 homodimers and induced homodimerization of recombinant PD-L1. Our results suggest that pranlukast inhibits the PD-1/PD-L1 axis, meriting its repurposing as an antitumor drug.

Nucleotide analogues and mpox: Repurposing the repurposable.

While the COVID-19 crisis is still ongoing, a new public health threat has emerged with recent outbreaks of monkeypox (mpox) infections in Africa. Mass vaccination is not currently recommended by the World Health Organization (WHO), and antiviral treatments are yet to be specifically approved for mpox, although existing FDA-approved drugs (Tecovirimat, Brincidofovir, and Cidofovir) may be used in severe cases or for immunocompromised patients. A first-line of defense is thus drug repurposing, which was heavily attempted against SARS-CoV-2 - albeit with limited success. This review focuses on nucleoside analogues as promising antiviral candidates for targeting of the viral DNA-dependent DNA polymerase. In contrast to broad-spectrum screening approaches employed for SARS-CoV-2, we emphasize the importance of understanding the structural specificity of viral polymerases for rational selection of potential candidates. By comparing DNA-dependent DNA polymerases with other viral polymerases, we highlight the unique features that influence the efficacy and selectivity of nucleoside analogues. These structural insights provide a framework for the preselection, repurposing, optimization, and design of nucleoside analogues, aiming to accelerate the development of targeted antiviral therapies for mpox and other viral infections.

Exogenous binding immunoglobulin protein (BiP) enhance immune regulatory phenotype in ex-vivo Mtb infected PBMCs stratified based on QuantiFERON response.

Even though anti-tuberculosis (TB) treatment is readily available, Mycobacterium tuberculosis (Mtb) infection continues to be a global threat with a high death rate recorded from a single infectious agent. This highlights the significance of developing new strategies to curb the growing Mtb infection cases. Host-directed therapies (HDT) offer a promising approach that includes both drug discovery and drug repurposing, aimed at identifying host targets and promoting immune cell populations that can lead to better infection outcomes. In this context, we investigated the potential of exogenous Binding Immunoglobulin Protein (BiP) to induce such changes ex-vivo using PBMCs from healthy (QFN-) and Mtb exposed (QFN+) individuals. We analysed cell surface expression and cytokine profiles across eight different stimulation conditions including human full-length BiP protein (20μg/ml), TLR-9a (0.5μM), BiP/TLR-9a combination, isoniazid (1μM), H37Rv (MOI: 1: 10), and pooled bronchoalveolar lavage (BAL) samples collected at TB diagnosis (TBdx) and at month 6 (M6) of anti-TB treatment. Our results revealed that BiP-stimulated PBMCs showed a significant reduction of interleukin (IL)-10 secretion, along with increased IL-4, IL-5, IL-13, and soluble Fas-L (sFasL) secretion. We also observed that BiP stimulation enhanced the expression of membrane bound Fas-L (CD178) and IL5Ra (CD125) in B-cells isolated from both QFN- and QFN+ groups. Additionally, BiP exhibited a synergistic effect with TLR-9a, further boosting this co-expression. Moreover, we observed that BiP induced IL5Ra expression in both CD3+CD5lo and CD3+CD5hi T-cell populations. This study explores the effects of exogenous BiP on cell functionality and provides valuable insights into its potential to modulate host cell responses, which could be explored as a host-directed therapy for TB in the future.

Repurposing bosentan as an anticancer agent: EGFR/ERK/c-Jun modulation inhibits NSCLC tumor growth.

Drug repurposing of well-established drugs to be targeted against lung cancer has been a promising strategy. Bosentan is an endothelin 1 (ET-1) blocker widely used in pulmonary hypertension. The current experiment intends to inspect the anticancer and antiangiogenic mechanism of bosentan targeting epidermal growth factor receptor (EGFR) /extra-cellular Signal Regulated Kinase (ERK) /c-Jun/vascular endothelial growth factor (VEGF) carcinogenic pathway. BALB/c mice were randomized into four groups, the first received the vehicle, the second received 100 mg/kg oral bosentan alone, the third has non-small cell lung cancer (NSCLC) induced by two doses of 1.5g/kg urethane i.p. and finally the fourth has NSCLC received bosentan. To determine the anti-proliferative impact of bosentan, cytokeratin 19 fragments (CYFRA 21-1) level was assessed, and Ki-67 positive cells were counted by immunohistochemical (IHC). Molecular expression of EGFR via IHC, relative expression of p-ERK1/2 and p-c-Jun via western blotting and caspase 3, Bcl-2 Associated X-protein (BAX)/B-cell lymphoma 2 (Bcl-2) ratio and VEGF via ELISA were quantified. Bosentan showed pronounced improvement in lung index and histopathological examinations. Bosentan exerted a noticeable arrest of lung cancer growth indicated by the attenuation of CYFRA 21-1 and Ki-67 positive cell counts besides the boost of BAX/Bcl-2 ratio and caspase 3. Bosentan induced a remarkable decline of EGFR, T-ERK1/2/p-ERK1/2, T-c-Jun/p-c-Jun, and VEGF. Bosentan induced cytotoxic and anti-angiogenic impact through regulation of EGFR/ERK/c-Jun/VEGF axis suggesting its potential therapeutic impact against lung cancer.

Effect of long and short half-life PDE5 inhibitors on HbA1c levels: a systematic review and meta-analysis.

Phosphodiesterase 5 (PDE5) inhibitors, owing to their mechanism of action, have been gaining recognition as a potential case of drug repurposing and combination therapy for diabetes treatment. We aimed to examine the effect of long and short half-life PDE5 inhibitors have on Haemoglobin A1c (HbA1c) levels. A systematic review and meta-analysis was conducted of randomised controlled trials (RCTs) in people with elevated HbA1c (>6%) to assess mean difference in HbA1c levels from baseline versus controls after any PDE5 inhibitor intervention of ≥4 weeks, excluding multiple interventions. Cochrane CENTRAL, PMC Medline, ClinicalTrials.gov, and WHO ICTRP were searched without language restrictions up to September 30, 2024. Summary data from published data were extracted. PRISMA and Cochrane guidelines used to extract and assess data using a random-effects meta-analysis. This study is registered with the Research Registry, reviewregistry1733. Among 1096 studies identified, in analysis of 13 studies with 1083 baseline patients, long half-life PDE5 inhibitors (tadalafil, PF-00489791) had decreases in HbA1c while short half-life PDE5 inhibitors (sildenafil, avanafil) had no change. Five (38.5%) studies had a low risk of bias, and eight (61.5%) had some concerns. Long half-life inhibitors had significant mean decrease of-0.40% ([-0.66,-0.14], p=0.002, I2=82%, 7.70% baseline HbA1c). Short half-life inhibitors had insignificant mean difference of+0.08% ([-0.16, 0.33], p=0.51, I2=40%, 7.73% baseline HbA1c). In ≥8-week trials with participants with type 2 diabetes (T2D) and mean HbA1c≥6.5%, long half-life inhibitors had significant mean decrease of-0.50% ([-0.83,-0.17], I2=88%, p=0.003); short half-life inhibitors had significant mean increase of+0.36% ([0.03, 0.68], I2=3%, p=0.03). At the well-controlled HbA1c of the participants, previous literature shows current diabetes treatments have similar HbA1c decreases, so the HbA1c mean difference of long half-life PDE5 inhibitors may indeed be clinically relevant. This suggests future investigation into PDE5 inhibitors as part of combination therapy or as therapy for high HbA1c individuals is needed, especially because of variable risk of biases, homogeneity, and sample sizes in our study. None.

Iron metabolism dysfunction in neuropsychiatric disorders: Implications for therapeutic intervention.

Iron is a trace metal that takes part in the maintenance of body homeostasis by, for instance, aiding in energy production and immunity. A body of evidence now demonstrates that dysfunction in iron metabolism can have detrimental effects and is intricately associated with the development of neuropsychiatric disorders, including Major Depressive Disorder (MDD), anxiety, and schizophrenia. For instance, changes in serum and central nervous system (CNS) levels of iron and in proteins mediating iron metabolism have been documented in patients grappling with the aforementioned diseases. By contrast, targeting iron metabolism by using iron chelators, for instance, has proven to be effective in alleviating disease burden. Therefore, here we review the state-of-the-art regarding the role of iron metabolism and its dysfunction in the context of neuropsychiatric disorders. Furthermore, we discuss how targeting iron metabolism can be an effective therapeutic option to tackle this class of diseases. Finally, we discuss the mechanisms linking this dysfunction to behavioral changes in these disorders. Harnessing the knowledge of iron metabolism is not only key to the characterization of novel molecular targets and disease biomarkers but also crucial to drug repurposing and drug design.

Abstract TP211: In silico drug repurposing identifies transcriptional mitigators of blood-induced neurotoxicity in an organoid model of intracerebral hemorrhage

Abstract TP211: <i>In silico</i> drug repurposing identifies transcriptional mitigators of blood-induced neurotoxicity in an organoid model of intracerebral hemorrhage

Study on Drug Repurposing for ALS Treatment Using Pre-trained Knowledge Graph Embeddings: Methods and Findings

In this study, research has been conducted using pre-trained knowledge graph embedding for drug repurposing in treating ALS (Amyotrophic Lateral Sclerosis), and its results have been presented. Drug repurposing studies for ALS have been carried out through two main methods: disease-drug relationship and genes-drugs relationship. Drug repurposing recommendations for ALS have been provided by predicting connections between disease and drug entities on the DRKG (Drug Repurposing Knowledge Graph). The findings obtained from the study have been evaluated by comparing them with the list of clinical trial drugs obtained from Drugbank. DRKG has been utilized as a comprehensive biological knowledge graph containing genes, compounds, diseases, biological processes, side effects, and symptoms. This graph has proven to be an effective resource for extracting information related to ALS disease. In drug repurposing studies, drugs obtained through disease-drug relationships have been compared with the list of clinical trial drugs associated with ALS, yielding significant results. Additionally, interactions between genes associated with ALS and drugs related to these genes have been examined in studies conducted through gene-drug relationships. The results obtained from the study demonstrate that DRKG is an effective resource for identifying drugs with potential therapeutic effects in the treatment of ALS. The findings suggest that drug repurposing studies could offer new and effective solutions for the treatment of ALS, marking a significant step forward in this regard.

Multi-ancestry genome-wide meta-analysis with 472,819 individuals identifies 32 novel risk loci for psoriasis.

Psoriasis is a common chronic, recurrent, immune-mediated disease involved in the skin or joints or both. However, deeper insight into the genetic susceptibility of psoriasis is still unclear. Here we performed the largest multi-ancestry meta-analysis of genome-wide association study including 28,869 psoriasis cases and 443,950 healthy controls. We identified 74 genome-wide significant loci for psoriasis. Of 74 loci, 32 were novel psoriasis risk loci. Across 74 loci, 801 likely causal genes are indicated and 164 causal genes are prioritized. SNP-based heritability analyses demonstrated that common variants explain 15% of genetic risk for psoriasis. Gene-set analyses and the genetic correlation revealed that psoriasis-related genes have the positive correlations with autoimmune diseases such as ulcerative colitis, inflammatory bowel diseases, and Crohn's disease. Gene-drug interaction analysis suggested that psoriasis-associated genes overlapped with targets of current medications for psoriasis. Finally, we used the multi-ancestry meta-analysis to explore drug repurposing and the potential targets for psoriasis. We identified 74 genome-wide significant loci for psoriasis. Based on 74 loci, we provided new biological insights to the etiology of psoriasis. Of clinical interest, we gave some hints for 76 potential targets and drug repurposing for psoriasis.

Scaled and Weighted Laplacian Matrices as Functional Descriptors for GPCR Ligands.

The G protein-coupled receptor (GPCR) pharmacology accounts for a significant field in research, clinical studies, and therapeutics. Computer-aided drug discovery is an evolving suite of techniques and methodologies that facilitate accelerated progress in drug discovery and repositioning. However, the structure-activity relationships of molecules targeting GPCRs are highly challenging in many cases since slight structural modifications can lead to drastic changes in biological functionality. Numerous molecular descriptors have been described, many of which successfully characterize the structural and physicochemical features of drug sets. Nonetheless, elucidating the structure-functionality relationships over extensive sets of drugs with multiple structural variations and known biological activity remains challenging in various biological systems. This work presents novel topological descriptors using Laplacian matrices, weighted, and scaled by atomic mass and partial charges. We tested these descriptors on three sets of GPCR ligands: muscarinic, β-adrenergic, and δ-opioid receptor ligands, evaluating their potential as functional descriptors of these receptors.

Identification of Potential Therapeutic Targets for Sensorineural Hearing Loss and Evaluation of Drug Development Potential Using Mendelian Randomization Analysis

Background: Sensorineural hearing loss (SNHL) is a major contributor to hearing impairment, yet effective therapeutic options remain elusive. Mendelian randomization (MR) has proven valuable for drug repurposing and identifying new therapeutic targets. This study aims to pinpoint novel treatment targets for SNHL, exploring their pathophysiological roles and potential adverse effects. Methods: This research utilized the UKB-PPP database to access cis-protein quantitative trait locus (cis-pQTL) data, with SNHL data sourced from the FinnGen database as the endpoint for the MR causal analysis of drug targets. Colocalization analysis was employed to determine whether SNHL risk and protein expression share common SNPs. A phenotype-wide association analysis was conducted to assess the potential side effects of these targets. Drug prediction and molecular docking were subsequently used to evaluate the therapeutic potential of the identified targets. Results: Four drug target proteins significantly associated with sensorineural hearing loss (SNHL) were determined by Mendelian randomization (MR) analysis and co-localization analysis. These drug targets include LATS1, TEF, LMNB2, and OGFR and were shown to have fewer potential side effects when acting on these target proteins by phenotype-wide association analysis. Genes associated with sensorineural hearing loss are primarily implicated in the Hippo signaling pathway, cell–cell adhesion, and various binding regulatory activities and are involved in the regulation of cell proliferation and apoptosis. Next, drugs for the treatment of SNHL were screened by the DsigDB database and molecular docking, and the top 10 drugs were selected based on p-value. Among them, atrazine CTD 00005450 was identified as the most likely therapeutic target, followed by ampyrone HL60 DOWN and genistein CTD 00007324. In addition, LMNB2, LATS1, and OGFR could be intervened in by multiple drugs; however, fewer drugs intervened in TEF. Conclusion: This study has successfully identified four promising drug targets for SNHL, which are likely to be effective in clinical trials with minimal side effects. These findings could significantly streamline drug development for SNHL, potentially reducing the costs and time associated with pharmaceutical research and development.

Mapping knowledge landscapes and emerging trends in artificial intelligence for antimicrobial resistance: bibliometric and visualization analysis

ObjectiveTo systematically map the knowledge landscape and development trends in artificial intelligence (AI) applications for antimicrobial resistance (AMR) research through bibliometric analysis, providing evidence-based insights to guide future research directions and inform strategic decision-making in this dynamic field.MethodsA comprehensive bibliometric analysis was performed using the Web of Science Core Collection database for publications from 2014 to 2024. The analysis integrated multiple bibliometric approaches: VOSviewer for visualization of collaboration networks and research clusters, CiteSpace for temporal evolution analysis, and quantitative analysis of publication metrics. Key bibliometric indicators including co-authorship patterns, keyword co-occurrence, and citation impact were analyzed to delineate research evolution and collaboration patterns in this domain.ResultsA collection of 2,408 publications was analyzed, demonstrating significant annual growth with publications increasing from 4 in 2014 to 549 in 2023 (22.7% of total output). The United States (707), China (581), and India (233) were the leading contributors in international collaborations. The Chinese Academy of Sciences (53), Harvard Medical School (43), and University of California San Diego (26) were identified as top contributing institutions. Citation analysis highlighted two major breakthroughs: AlphaFold’s protein structure prediction (6,811 citations) and deep learning approaches to antibiotic discovery (4,784 citations). Keyword analysis identified six enduring research clusters from 2014 to 2024: sepsis, artificial neural networks, antimicrobial resistance, antimicrobial peptides, drug repurposing, and molecular docking, demonstrating the sustained integration of AI in antimicrobial therapy development. Recent trends show increasing application of AI technologies in traditional approaches, particularly in MALDI-TOF MS for pathogen identification and graph neural networks for large-scale molecular screening.ConclusionThis bibliometric analysis shows the importance of artificial intelligence in enhancing the progress in the discovery of antimicrobial drugs especially toward the fight against AMR. From enhancing the fast, efficient and predictive performance of drug discovery methods, current AI capabilities have revealed observable potential to be proactive in combating the ever-growing challenge of AMR worldwide. This study serves not only an identification of current trends, but also, and especially, offers a strategic approach to further investigations.

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.

IAP dependency of T-cell prolymphocytic leukemia identified by high-throughput drug screening.

T-cell prolymphocytic leukemia (T-PLL) is an aggressive lymphoid malignancy with limited treatment options. To discover new treatment targets for T-PLL, we performed high-throughput drug sensitivity screening on 30 primary patient samples ex-vivo. After screening over 2'800 unique compounds, we found T-PLL to be more resistant to most drug classes, including chemotherapeutics, compared to other blood cancers. Furthermore, we discovered previously not reported vulnerabilities of T-PLL. T-PLL cells exhibited a particular sensitivity to drugs targeting autophagy (thapsigargin, bafilomycin A1), nuclear export (selinexor), and inhibitor of apoptosis proteins (IAPs) (birinapant), sensitivities that were also shared by other T-cell malignancies. Through bulk and single-cell RNA-Sequencing we found these compounds to activate the toll-like-receptor (TLR) (bafilomycin A1), p53 (selinexor), and TNF-ɑ/NFκB signaling pathways (birinapant) in T-PLL cells. Focussing on birinapant for its potential in drug repurposing, we uncovered that IAP inhibitor-induced cell death was primarily necroptotic and dependent on TNF-ɑ. Through spectral flow cytometry we confirmed the absence of cleaved caspase-3 in IAP inhibitor treated T-PLL cells and show that IAP inhibition reduces the proliferation of T-PLL cells stimulated ex-vivo, while showing only a limited effect on non-malignant T-cells. In summary, our study maps the drug sensitivity of T-PLL across a broad range of targets and identifies new therapeutic approaches for T-PLL by targeting IAPs, XPO1 and autophagy, highlighting potential candidates for drug repurposing and novel treatment strategies.

Storage and Query of Drug Knowledge Graphs Using Distributed Graph Databases: A Case Study

Background: Distributed graph databases are a promising method for storing and conducting complex pathway queries on large-scale drug knowledge graphs to support drug research. However, there is a research gap in evaluating drug knowledge graphs’ storage and query performance based on distributed graph databases. This study evaluates the feasibility and performance of distributed graph databases in managing large-scale drug knowledge graphs. Methods: First, a drug knowledge graph storage and query system is designed based on the Nebula Graph database. Second, the system’s writing and query performance is evaluated. Finally, two drug repurposing benchmarks are used to provide a more extensive and reliable assessment. Results: The performance of distributed graph databases surpasses that of single-machine databases, including data writing, regular queries, constrained queries, and concurrent queries. Additionally, the advantages of distributed graph databases in writing performance become more pronounced as the data volume increases. The query performance benefits of distributed graph databases also improve with the complexity of query tasks. The drug repurposing evaluation results show that 78.54% of the pathways are consistent with currently approved drug treatments according to repoDB. Additionally, 12 potential pathways for new drug indications are found to have literature support according to DrugRepoBank. Conclusions: The proposed system is able to construct, store, and query a large graph of multisource drug knowledge and provides reliable and explainable drug–disease paths for drug repurposing.