Multi-view feature representation and fusion for drug-drug interactions prediction

A systematic review of molecular structures, knowledge graphs, and cold-start scenario in drug-drug interaction prediction

Drugs are essential chemical substances used to treat, prevent, or diagnose diseases. However, the co-administration or simultaneous use of multiple drugs can lead to complex interactions known as Drug-Drug Interactions (DDIs). Adverse DDIs are a major concern in the pharmaceutical industry, posing significant health risks by potentially causing toxicity or negatively impacting therapeutic efficacy. Identifying all DDIs through clinical trials alone is not feasible, making computational approaches crucial for predicting and understanding these interactions. Molecular structures and biomedical entities represented as knowledge graphs (KGs) have been widely used for DDI prediction. This review first examines diverse molecular representations commonly utilized in DDI prediction. It then highlights KG-based approaches, emphasizing their ability to integrate heterogeneous biomedical data and provide comprehensive structural and relational insights into drugs, proteins, and other biological entities. However, predictive models frequently encounter challenges when dealing with drugs with limited interaction data or unknown structures, resulting in a cold-start scenario that negatively impacts model generalization. Consequently, this review also discusses methods to address the cold-start scenario in DDI prediction. Finally, key findings and potential directions for enhancing DDI prediction are presented.

Artificial Intelligence Models and Tools for the Assessment of Drug-Herb Interactions

Artificial intelligence (AI) has emerged as a powerful tool in medical sciences that is revolutionizing various fields of drug research. AI algorithms can analyze large-scale biological data and identify molecular targets and pathways advancing pharmacological knowledge. An especially promising area is the assessment of drug interactions. The AI analysis of large datasets, such as drugs' chemical structure, pharmacological properties, molecular pathways, and known interaction patterns, can provide mechanistic insights and identify potential associations by integrating all this complex information and returning potential risks associated with these interactions. In this context, an area where AI may prove valuable is in the assessment of the underlying mechanisms of drug interactions with natural products (i.e., herbs) that are used as dietary supplements. These products pose a challenging problem since they are complex mixtures of constituents with diverse and limited information regarding their pharmacological properties, especially their pharmacokinetic data. As the use of herbal products and supplements continues to grow, it becomes increasingly important to understand the potential interactions between them and conventional drugs and the associated adverse drug reactions. This review will discuss AI approaches and how they can be exploited in providing valuable mechanistic insights regarding the prediction of interactions between drugs and herbs, and their potential exploitation in experimental validation or clinical utilization.

Curcumin nanoparticles in heat stroke management

OBJECTIVE

The exacerbation of extreme high-temperature events due to global climate change poses a significant challenge to public health, particularly impacting the central nervous system through heat stroke. This study aims to develop Poly(amidoamine) (PAMAM) nanoparticles loaded with curcumin (PAMAM@Cur) to enhance its therapeutic efficacy in hypothalamic neural damage in a heat stroke model and explore its potential mechanisms.

METHODS

Curcumin (Cur) was encapsulated into PAMAM nanoparticles through a hydrophobic interaction method, and various techniques were employed to characterize their physicochemical properties. A heat stroke mouse model was established to monitor body temperature and serum biochemical parameters, conduct behavioral assessments, histological examinations, and biochemical analyses. Transcriptomic and proteomic analyses were performed to investigate the therapeutic mechanisms of PAMAM@Cur, validated in an N2a cell model.

RESULTS

PAMAM@Cur demonstrated good stability, photostability, cell compatibility, significant blood-brain barrier (BBB) penetration capability, and effective accumulation in the brain. PAMAM@Cur markedly improved behavioral performance and neural cell structural integrity in heat stroke mice, alleviated inflammatory responses, with superior therapeutic effects compared to Cur or PAMAM alone. Multi-omics analysis revealed that PAMAM@Cur regulated antioxidant defense genes and iron death-related genes, particularly upregulating the PCBP2 protein, stabilizing SLC7A11 and GPX4 mRNA, and reducing iron-dependent cell death.

CONCLUSION

By enhancing the drug delivery properties of Cur and modulating molecular pathways relevant to disease treatment, PAMAM@Cur significantly enhances the therapeutic effects against hypothalamic neural damage induced by heat stroke, showcasing the potential of nanotechnology in improving traditional drug efficacy and providing new strategies for future clinical applications.

SIGNIFICANCE

This study highlights the outlook of nanotechnology in treating neurological disorders caused by heat stroke, offering a novel therapeutic approach with potential clinical applications.

Prediction of multiple types of drug interactions based on multi-scale fusion and dual-view fusion

Potential drug-drug interactions (DDI) can lead to adverse drug reactions (ADR), and DDI prediction can help pharmacy researchers detect harmful DDI early. However, existing DDI prediction methods fall short in fully capturing drug information. They typically employ a single-view input, focusing solely on drug features or drug networks. Moreover, they rely exclusively on the final model layer for predictions, overlooking the nuanced information present across various network layers. To address these limitations, we propose a multi-scale dual-view fusion (MSDF) method for DDI prediction. More specifically, MSDF first constructs two views, topological and feature views of drugs, as model inputs. Then a graph convolutional neural network is used to extract the feature representations from each view. On top of that, a multi-scale fusion module integrates information across different graph convolutional layers to create comprehensive drug embeddings. The embeddings from the two views are summed as the final representation for classification. Experiments on two real-world datasets demonstrate that MSDF achieves higher accuracy than state-of-the-art methods, as the dual-view, multi-scale approach better captures drug characteristics.

Attention is all you need: utilizing attention in AI-enabled drug discovery

Recently, attention mechanism and derived models have gained significant traction in drug development due to their outstanding performance and interpretability in handling complex data structures. This review offers an in-depth exploration of the principles underlying attention-based models and their advantages in drug discovery. We further elaborate on their applications in various aspects of drug development, from molecular screening and target binding to property prediction and molecule generation. Finally, we discuss the current challenges faced in the application of attention mechanisms and Artificial Intelligence technologies, including data quality, model interpretability and computational resource constraints, along with future directions for research. Given the accelerating pace of technological advancement, we believe that attention-based models will have an increasingly prominent role in future drug discovery. We anticipate that these models will usher in revolutionary breakthroughs in the pharmaceutical domain, significantly accelerating the pace of drug development.

HTCL-DDI: a hierarchical triple-view contrastive learning framework for drug-drug interaction prediction

Drug-drug interaction (DDI) prediction can discover potential risks of drug combinations in advance by detecting drug pairs that are likely to interact with each other, sparking an increasing demand for computational methods of DDI prediction. However, existing computational DDI methods mostly rely on the single-view paradigm, failing to handle the complex features and intricate patterns of DDIs due to the limited expressiveness of the single view. To this end, we propose a Hierarchical Triple-view Contrastive Learning framework for Drug-Drug Interaction prediction (HTCL-DDI), leveraging the molecular, structural and semantic views to model the complicated information involved in DDI prediction. To aggregate the intra-molecular compositional and structural information, we present a dual attention-aware network in the molecular view. Based on the molecular view, to further capture inter-molecular information, we utilize the one-hop neighboring information and high-order semantic relations in the structural view and semantic view, respectively. Then, we introduce contrastive learning to enhance drug representation learning from multifaceted aspects and improve the robustness of HTCL-DDI. Finally, we conduct extensive experiments on three real-world datasets. All the experimental results show the significant improvement of HTCL-DDI over the state-of-the-art methods, which also demonstrates that HTCL-DDI opens new avenues for ensuring medication safety and identifying synergistic drug combinations.