A dual-modal graph learning framework for identifying interaction events among chemical and biotech drugs.

Drug-drug interaction (DDI) identification is essential to clinical medicine and drug discovery. The two categories of drugs (i.e. chemical drugs and biotech drugs) differ remarkably in molecular properties, action mechanisms, etc. Biotech drugs are up-to-comers but highly promising in modern medicine due to higher specificity and fewer side effects. However, existing DDI prediction methods only consider chemical drugs of small molecules, not biotech drugs of large molecules. Here, we build a large-scale dual-modal graph database named CB-DB and customize a graph-based framework named CB-TIP to reason event-aware DDIs for both chemical and biotech drugs. CB-DB comprehensively integrates various interaction events and two heterogeneous kinds of molecular structures. It imports endogenous proteins founded on the fact that most drugs take effects by interacting with endogenous proteins. In the modality of molecular structure, drugs and endogenous proteins are two heterogeneous kinds of graphs, while in the modality of interaction, they are nodes connected by events (i.e. edges of different relationships). CB-TIP employs graph representation learning methods to generate drug representations from either modality and then contrastively mixes them to predict how likely an event occurs when a drug meets another in an end-to-end manner. Experiments demonstrate CB-TIP's great superiority in DDI prediction and the promising potential of uncovering novel DDIs.

R2-DDI: relation-aware feature refinement for drug-drug interaction prediction.

Precisely predicting the drug-drug interaction (DDI) is an important application and host research topic in drug discovery, especially for avoiding the adverse effect when using drug combination treatment for patients. Nowadays, machine learning and deep learning methods have achieved great success in DDI prediction. However, we notice that most of the works ignore the importance of the relation type when building the DDI prediction models. In this work, we propose a novel R$^2$-DDI framework, which introduces a relation-aware feature refinement module for drug representation learning. The relation feature is integrated into drug representation and refined in the framework. With the refinement features, we also incorporate the consistency training method to regularize the multi-branch predictions for better generalization. Through extensive experiments and studies, we demonstrate our R$^2$-DDI approach can significantly improve the DDI prediction performance over multiple real-world datasets and settings, and our method shows better generalization ability with the help of the feature refinement design.

Attention-based cross domain graph neural network for prediction of drug-drug interactions.

Drug-drug interactions (DDI) may lead to adverse reactions in human body and accurate prediction of DDI can mitigate the medical risk. Currently, most of computer-aided DDI prediction methods construct models based on drug-associated features or DDI network, ignoring the potential information contained in drug-related biological entities such as targets and genes. Besides, existing DDI network-based models could not make effective predictions for drugs without any known DDI records. To address the above limitations, we propose an attention-based cross domain graph neural network (ACDGNN) for DDI prediction, which considers the drug-related different entities and propagate information through cross domain operation. Different from the existing methods, ACDGNN not only considers rich information contained in drug-related biomedical entities in biological heterogeneous network, but also adopts cross-domain transformation to eliminate heterogeneity between different types of entities. ACDGNN can be used in the prediction of DDIs in both transductive and inductive setting. By conducting experiments on real-world dataset, we compare the performance of ACDGNN with several state-of-the-art methods. The experimental results show that ACDGNN can effectively predict DDIs and outperform the comparison models.

Drug-drug interaction prediction: databases, web servers and computational models.

In clinical treatment, two or more drugs (i.e. drug combination) are simultaneously or successively used for therapy with the purpose of primarily enhancing the therapeutic efficacy or reducing drug side effects. However, inappropriate drug combination may not only fail to improve efficacy, but even lead to adverse reactions. Therefore, according to the basic principle of improving the efficacy and/or reducing adverse reactions, we should study drug-drug interactions (DDIs) comprehensively and thoroughly so as to reasonably use drug combination. In this review, we first introduced the basic conception and classification of DDIs. Further, some important publicly available databases and web servers about experimentally verified or predicted DDIs were briefly described. As an effective auxiliary tool, computational models for predicting DDIs can not only save the cost of biological experiments, but also provide relevant guidance for combination therapy to some extent. Therefore, we summarized three types of prediction models (including traditional machine learning-based models, deep learning-based models and score function-based models) proposed during recent years and discussed the advantages as well as limitations of them. Besides, we pointed out the problems that need to be solved in the future research of DDIs prediction and provided corresponding suggestions.

DSN-DDI: an accurate and generalized framework for drug-drug interaction prediction by dual-view representation learning.

Drug-drug interaction (DDI) prediction identifies interactions of drug combinations in which the adverse side effects caused by the physicochemical incompatibility have attracted much attention. Previous studies usually model drug information from single or dual views of the whole drug molecules but ignore the detailed interactions among atoms, which leads to incomplete and noisy information and limits the accuracy of DDI prediction. In this work, we propose a novel dual-view drug representation learning network for DDI prediction ('DSN-DDI'), which employs local and global representation learning modules iteratively and learns drug substructures from the single drug ('intra-view') and the drug pair ('inter-view') simultaneously. Comprehensive evaluations demonstrate that DSN-DDI significantly improved performance on DDI prediction for the existing drugs by achieving a relatively improved accuracy of 13.01% and an over 99% accuracy under the transductive setting. More importantly, DSN-DDI achieves a relatively improved accuracy of 7.07% to unseen drugs and shows the usefulness for real-world DDI applications. Finally, DSN-DDI exhibits good transferability on synergistic drug combination prediction and thus can serve as a generalized framework in the drug discovery field.

Improving drug-drug interactions prediction with interpretability via meta-path-based information fusion.

Drug-drug interactions (DDIs) are compound effects when patients take two or more drugs at the same time, which may weaken the efficacy of drugs or cause unexpected side effects. Thus, accurately predicting DDIs is of great significance for the drug development and the drug safety surveillance. Although many methods have been proposed for the task, the biological knowledge related to DDIs is not fully utilized and the complex semantics among drug-related biological entities are not effectively captured in existing methods, leading to suboptimal performance. Moreover, the lack of interpretability for the predicted results also limits the wide application of existing methods for DDIs prediction. In this study, we propose a novel framework for predicting DDIs with interpretability. Specifically, we construct a heterogeneous information network (HIN) by explicitly utilizing the biological knowledge related to the procedure of inducing DDIs. To capture the complex semantics in HIN, a meta-path-based information fusion mechanism is proposed to learn high-quality representations of drugs. In addition, an attention mechanism is designed to combine semantic information obtained from meta-paths with different lengths to obtain final representations of drugs for DDIs prediction. Comprehensive experiments are conducted on 2410 approved drugs, and the results of predictive performance comparison show that our proposed framework outperforms selected representative baselines on the task of DDIs prediction. The results of ablation study and cold-start scenario indicate that the meta-path-based information fusion mechanism red is beneficial for capturing the complex semantics among drug-related biological entities. Moreover, the results of case study demonstrate that the designed attention mechanism is able to provide partial interpretability for the predicted DDIs. Therefore, the proposed method will be a feasible solution to the task of predicting DDIs.

A social theory-enhanced graph representation learning framework for multitask prediction of drug-drug interactions.

Current machine learning-based methods have achieved inspiring predictions in the scenarios of mono-type and multi-type drug-drug interactions (DDIs), but they all ignore enhancive and depressive pharmacological changes triggered by DDIs. In addition, these pharmacological changes are asymmetric since the roles of two drugs in an interaction are different. More importantly, these pharmacological changes imply significant topological patterns among DDIs. To address the above issues, we first leverage Balance theory and Status theory in social networks to reveal the topological patterns among directed pharmacological DDIs, which are modeled as a signed and directed network. Then, we design a novel graph representation learning model named SGRL-DDI (social theory-enhanced graph representation learning for DDI) to realize the multitask prediction of DDIs. SGRL-DDI model can capture the task-joint information by integrating relation graph convolutional networks with Balance and Status patterns. Moreover, we utilize task-specific deep neural networks to perform two tasks, including the prediction of enhancive/depressive DDIs and the prediction of directed DDIs. Based on DDI entries collected from DrugBank, the superiority of our model is demonstrated by the comparison with other state-of-the-art methods. Furthermore, the ablation study verifies that Balance and Status patterns help characterize directed pharmacological DDIs, and that the joint of two tasks provides better DDI representations than individual tasks. Last, we demonstrate the practical effectiveness of our model by a version-dependent test, where 88.47 and 81.38% DDI out of newly added entries provided by the latest release of DrugBank are validated in two predicting tasks respectively.

DMFDDI: deep multimodal fusion for drug-drug interaction prediction.

Drug combination therapy has gradually become a promising treatment strategy for complex or co-existing diseases. As drug-drug interactions (DDIs) may cause unexpected adverse drug reactions, DDI prediction is an important task in pharmacology and clinical applications. Recently, researchers have proposed several deep learning methods to predict DDIs. However, these methods mainly exploit the chemical or biological features of drugs, which is insufficient and limits the performances of DDI prediction. Here, we propose a new deep multimodal feature fusion framework for DDI prediction, DMFDDI, which fuses drug molecular graph, DDI network and the biochemical similarity features of drugs to predict DDIs. To fully extract drug molecular structure, we introduce an attention-gated graph neural network for capturing the global features of the molecular graph and the local features of each atom. A sparse graph convolution network is introduced to learn the topological structure information of the DDI network. In the multimodal feature fusion module, an attention mechanism is used to efficiently fuse different features. To validate the performance of DMFDDI, we compare it with 10 state-of-the-art methods. The comparison results demonstrate that DMFDDI achieves better performance in DDI prediction. Our method DMFDDI is implemented in Python using the Pytorch machine-learning library, and it is freely available at https://github.com/DHUDEBLab/DMFDDI.git.

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.

HetDDI: a pre-trained heterogeneous graph neural network model for drug-drug interaction prediction.

The simultaneous use of two or more drugs due to multi-disease comorbidity continues to increase, which may cause adverse reactions between drugs that seriously threaten public health. Therefore, the prediction of drug-drug interaction (DDI) has become a hot topic not only in clinics but also in bioinformatics. In this study, we propose a novel pre-trained heterogeneous graph neural network (HGNN) model named HetDDI, which aggregates the structural information in drug molecule graphs and rich semantic information in biomedical knowledge graph to predict DDIs. In HetDDI, we first initialize the parameters of the model with different pre-training methods. Then we apply the pre-trained HGNN to learn the feature representation of drugs from multi-source heterogeneous information, which can more effectively utilize drugs' internal structure and abundant external biomedical knowledge, thus leading to better DDI prediction. We evaluate our model on three DDI prediction tasks (binary-class, multi-class and multi-label) with three datasets and further assess its performance on three scenarios (S1, S2 and S3). The results show that the accuracy of HetDDI can achieve 98.82% in the binary-class task, 98.13% in the multi-class task and 96.66% in the multi-label one on S1, which outperforms the state-of-the-art methods by at least 2%. On S2 and S3, our method also achieves exciting performance. Furthermore, the case studies confirm that our model performs well in predicting unknown DDIs. Source codes are available at https://github.com/LinsLab/HetDDI.

A model-agnostic framework to enhance knowledge graph-based drug combination prediction with drug-drug interaction data and supervised contrastive learning.

Combination therapies have brought significant advancements to the treatment of various diseases in the medical field. However, searching for effective drug combinations remains a major challenge due to the vast number of possible combinations. Biomedical knowledge graph (KG)-based methods have shown potential in predicting effective combinations for wide spectrum of diseases, but the lack of credible negative samples has limited the prediction performance of machine learning models. To address this issue, we propose a novel model-agnostic framework that leverages existing drug-drug interaction (DDI) data as a reliable negative dataset and employs supervised contrastive learning (SCL) to transform drug embedding vectors to be more suitable for drug combination prediction. We conducted extensive experiments using various network embedding algorithms, including random walk and graph neural networks, on a biomedical KG. Our framework significantly improved performance metrics compared to the baseline framework. We also provide embedding space visualizations and case studies that demonstrate the effectiveness of our approach. This work highlights the potential of using DDI data and SCL in finding tighter decision boundaries for predicting effective drug combinations.

MCFF-MTDDI: multi-channel feature fusion for multi-typed drug-drug interaction prediction.

Adverse drug-drug interactions (DDIs) have become an increasingly serious problem in the medical and health system. Recently, the effective application of deep learning and biomedical knowledge graphs (KGs) have improved the DDI prediction performance of computational models. However, the problems of feature redundancy and KG noise also arise, bringing new challenges for researchers. To overcome these challenges, we proposed a Multi-Channel Feature Fusion model for multi-typed DDI prediction (MCFF-MTDDI). Specifically, we first extracted drug chemical structure features, drug pairs' extra label features, and KG features of drugs. Then, these different features were effectively fused by a multi-channel feature fusion module. Finally, multi-typed DDIs were predicted through the fully connected neural network. To our knowledge, we are the first to integrate the extra label information into KG-based multi-typed DDI prediction; besides, we innovatively proposed a novel KG feature learning method and a State Encoder to obtain target drug pairs' KG-based features which contained more abundant and more key drug-related KG information with less noise; furthermore, a Gated Recurrent Unit-based multi-channel feature fusion module was proposed in an innovative way to yield more comprehensive feature information about drug pairs, effectively alleviating the problem of feature redundancy. We experimented with four datasets in the multi-class and the multi-label prediction tasks to comprehensively evaluate the performance of MCFF-MTDDI for predicting interactions of known-known drugs, known-new drugs and new-new drugs. In addition, we further conducted ablation studies and case studies. All the results fully demonstrated the effectiveness of MCFF-MTDDI.

Predicting Drug-drug Interaction with Graph Mutual Interaction Attention Mechanism.

Effective representation of molecules is a crucial step in AI-driven drug design and drug discovery, especially for drug-drug interaction (DDIs) prediction. Previous work usually models the drug information from the drug-related knowledge graph or the single drug molecules, but the interaction information between molecular substructures of drug pair is seldom considered, thus often ignoring the influence of bond information on atom node representation, leading to insufficient drug representation. Moreover, key molecular substructures have significant contribution to the DDIs prediction results. Therefore, in this work, we propose a novel Graph learning framework of Mutual Interaction Attention mechanism (called GMIA) to predict DDIs by effectively representing the drug molecules. Specifically, we build the node-edge message communication encoder to aggregate atom node and the incoming edge information for atom node representation and design the mutual interaction attention decoder to capture the mutual interaction context between molecular graphs of drug pairs. GMIA can bridge the gap between two encoders for the single drug molecules by attention mechanism. We also design a co-attention matrix to analyze the significance of different-size substructures obtained from the encoder-decoder layer and provide interpretability. In comparison with other recent state-of-the-art methods, our GMIA achieves the best results in terms of area under the precision-recall-curve (AUPR), area under the ROC curve (AUC), and F1 score on two different scale datasets. The case study indicates that our GMIA can detect the key substructure for potential DDIs, demonstrating the enhanced performance and interpretation ability of GMIA.

The molecular basis of dapsone activation of CYP2C9-catalyzed nonsteroidal anti-inflammatory drug oxidation.

Positive heterotropic cooperativity, or "activation," results in an instantaneous increase in enzyme activity in the absence of an increase in protein expression. Thus, cytochrome P450 (CYP) enzyme activation presents as a potential drug-drug interaction mechanism. It has been demonstrated previously that dapsone activates the CYP2C9-catalyzed oxidation of a number of nonsteroidal anti-inflammatory drugs in vitro. Here, we conducted molecular dynamics simulations (MDS) together with enzyme kinetic investigations and site-directed mutagenesis to elucidate the molecular basis of the activation of CYP2C9-catalyzed S-flurbiprofen 4'-hydroxylation and S-naproxen O-demethylation by dapsone. Supplementation of incubations of recombinant CYP2C9 with dapsone increased the catalytic efficiency of flurbiprofen and naproxen oxidation by 2.3- and 16.5-fold, respectively. MDS demonstrated that activation arises predominantly from aromatic interactions between the substrate, dapsone, and the phenyl rings of Phe114 and Phe476 within a common binding domain of the CYP2C9 active site, rather than involvement of a distinct effector site. Mutagenesis of Phe114 and Phe476 abrogated flurbiprofen and naproxen oxidation, and MDS and kinetic studies with the CYP2C9 mutants further identified a pivotal role of Phe476 in dapsone activation. MDS additionally showed that aromatic stacking interactions between two molecules of naproxen are necessary for binding in a catalytically favorable orientation. In contrast to flurbiprofen and naproxen, dapsone did not activate the 4'-hydroxylation of diclofenac, suggesting that the CYP2C9 active site favors cooperative binding of nonsteroidal anti-inflammatory drugs with a planar or near-planar geometry. More generally, the work confirms the utility of MDS for investigating ligand binding in CYP enzymes.

Twice-Daily Dosing of Dolutegravir in Infants on Rifampicin Treatment: A Pharmacokinetic Substudy of the EMPIRICAL Trial.

BACKGROUND: We evaluated dolutegravir pharmacokinetics in infants with human immunodeficiency virus (HIV) receiving dolutegravir twice daily (BID) with rifampicin-based tuberculosis (TB) treatment compared with once daily (OD) without rifampicin. METHODS: Infants with HIV aged 1-12 months, weighing ≥3 kg, and receiving dolutegravir BID with rifampicin or OD without rifampicin were eligible. Six blood samples were taken over 12 (BID) or 24 hours (OD). Dolutegravir pharmacokinetic parameters, HIV viral load (VL) data, and adverse events (AEs) were reported. RESULTS: Twenty-seven of 30 enrolled infants had evaluable pharmacokinetic curves. The median (interquartile range) age was 7.1 months (6.1-9.9), weight was 6.3 kg (5.6-7.2), 21 (78%) received rifampicin, and 11 (41%) were female. Geometric mean ratios comparing dolutegravir BID with rifampicin versus OD without rifampicin were area under curve (AUC)0-24h 0.91 (95% confidence interval, .59-1.42), Ctrough 0.95 (0.57-1.59), Cmax 0.87 (0.57-1.33). One infant (5%) receiving rifampicin versus none without rifampicin had dolutegravir Ctrough <0.32 mg/L, and none had Ctrough <0.064 mg/L. The dolutegravir metabolic ratio (dolutegravir-glucuronide AUC/dolutegravir AUC) was 2.3-fold higher in combination with rifampicin versus without rifampicin. Five of 82 reported AEs were possibly related to rifampicin or dolutegravir and resolved without treatment discontinuation. Upon TB treatment completion, HIV viral load was <1000 copies/mL in 76% and 100% of infants and undetectable in 35% and 20% of infants with and without rifampicin, respectively. CONCLUSIONS: Dolutegravir BID in infants receiving rifampicin resulted in adequate dolutegravir exposure, supporting this treatment approach for infants with HIV-TB coinfection.

Assessing potential drug-drug interactions between clofazimine and other frequently used agents to treat drug-resistant tuberculosis.

Clofazimine is included in drug regimens to treat rifampicin/drug-resistant tuberculosis (DR-TB), but there is little information about its interaction with other drugs in DR-TB regimens. We evaluated the pharmacokinetic interaction between clofazimine and isoniazid, linezolid, levofloxacin, and cycloserine, dosed as terizidone. Newly diagnosed adults with DR-TB at Klerksdorp/Tshepong Hospital, South Africa, were started on the then-standard treatment with clofazimine temporarily excluded for the initial 2 weeks. Pharmacokinetic sampling was done immediately before and 3 weeks after starting clofazimine, and drug concentrations were determined using validated liquid chromatography-tandem mass spectrometry assays. The data were interpreted with population pharmacokinetics in NONMEM v7.5.1 to explore the impact of clofazimine co-administration and other relevant covariates on the pharmacokinetics of isoniazid, linezolid, levofloxacin, and cycloserine. Clofazimine, isoniazid, linezolid, levofloxacin, and cycloserine data were available for 16, 27, 21, 21, and 6 participants, respectively. The median age and weight for the full cohort were 39 years and 52 kg, respectively. Clofazimine exposures were in the expected range, and its addition to the regimen did not significantly affect the pharmacokinetics of the other drugs except levofloxacin, for which it caused a 15% reduction in clearance. A posteriori power size calculations predicted that our sample sizes had 97%, 90%, and 87% power at P < 0.05 to detect a 30% change in clearance of isoniazid, linezolid, and cycloserine, respectively. Although clofazimine increased the area under the curve of levofloxacin by 19%, this is unlikely to be of great clinical significance, and the lack of interaction with other drugs tested is reassuring.

Two-way pharmacodynamic modeling of drug combinations and its application to pairs of repurposed Ebola and SARS-CoV-2 agents.

Existing pharmacodynamic (PD) mathematical models for drug combinations discriminate antagonistic, additive, multiplicative, and synergistic effects, but fail to consider how concentration-dependent drug interaction effects may vary across an entire dose-response matrix. We developed a two-way pharmacodynamic (TWPD) model to capture the PD of two-drug combinations. TWPD captures interactions between upstream and downstream drugs that act on different stages of viral replication, by quantifying upstream drug efficacy and concentration-dependent effects on downstream drug pharmacodynamic parameters. We applied TWPD to previously published in vitro drug matrixes for repurposed potential anti-Ebola and anti-SARS-CoV-2 drug pairs. Depending on the drug pairing, the model recapitulated combined efficacies as or more accurately than existing models and can be used to infer efficacy at untested drug concentrations. TWPD fits the data slightly better in one direction for all drug pairs, meaning that we can tentatively infer the upstream drug. Based on its high accuracy, TWPD could be used in concert with PK models to estimate the therapeutic effects of drug pairs in vivo.

Assessing and mitigating pH-mediated DDI risks in drug development - formulation approaches and clinical considerations.

pH-mediated drug-drug interactions (DDI) is a prevalent DDI in drug development, especially for weak base compounds with highly pH-dependent solubility. FDA has released a guidance on the evaluation of pH-mediated DDI assessments using in vitro testing and clinical studies. Currently, there is no common practice of ways of testing across the academia and industry. The development of biopredictive method and physiologically-based biopharmaceutics modeling (PBBM) approaches to assess acid-reducing agent (ARA)-DDI have been proven with accurate prediction and could decrease drug development burden, inform clinical design and potentially waive clinical studies. Formulation strategies and careful clinical design could help mitigate the pH-mediated DDI to avoid more clinical studies and label restrictions, ultimately benefiting the patient. In this review paper, a detailed introduction on biorelevant dissolution testing, preclinical and clinical study requirement and PBPK modeling approaches to assess ARA-DDI are described. An improved decision tree for pH-mediated DDI is proposed. Potential mitigations including clinical or formulation strategies are discussed.

META-DDIE: predicting drug-drug interaction events with few-shot learning.

Drug-drug interactions (DDIs) are one of the major concerns in pharmaceutical research, and a number of computational methods have been developed to predict whether two drugs interact or not. Recently, more attention has been paid to events caused by the DDIs, which is more useful for investigating the mechanism hidden behind the combined drug usage or adverse reactions. However, some rare events may only have few examples, hindering them from being precisely predicted. To address the above issues, we present a few-shot computational method named META-DDIE, which consists of a representation module and a comparing module, to predict DDI events. We collect drug chemical structures and DDIs from DrugBank, and categorize DDI events into hundreds of types using a standard pipeline. META-DDIE uses the structures of drugs as input and learns the interpretable representations of DDIs through the representation module. Then, the model uses the comparing module to predict whether two representations are similar, and finally predicts DDI events with few labeled examples. In the computational experiments, META-DDIE outperforms several baseline methods and especially enhances the predictive capability for rare events. Moreover, META-DDIE helps to identify the key factors that may cause DDI events and reveal the relationship among different events.

Directed graph attention networks for predicting asymmetric drug-drug interactions.

It is tough to detect unexpected drug-drug interactions (DDIs) in poly-drug treatments because of high costs and clinical limitations. Computational approaches, such as deep learning-based approaches, are promising to screen potential DDIs among numerous drug pairs. Nevertheless, existing approaches neglect the asymmetric roles of two drugs in interaction. Such an asymmetry is crucial to poly-drug treatments since it determines drug priority in co-prescription. This paper designs a directed graph attention network (DGAT-DDI) to predict asymmetric DDIs. First, its encoder learns the embeddings of the source role, the target role and the self-roles of a drug. The source role embedding represents how a drug influences other drugs in DDIs. In contrast, the target role embedding represents how it is influenced by others. The self-role embedding encodes its chemical structure in a role-specific manner. Besides, two role-specific items, aggressiveness and impressionability, capture how the number of interaction partners of a drug affects its interaction tendency. Furthermore, the predictor of DGAT-DDI discriminates direction-specific interactions by the combination between two proximities and the above two role-specific items. The proximities measure the similarity between source/target embeddings and self-role embeddings. In the designated experiments, the comparison with state-of-the-art deep learning models demonstrates the superiority of DGAT-DDI across a direction-specific predicting task and a direction-blinded predicting task. An ablation study reveals how well each component of DGAT-DDI contributes to its ability. Moreover, a case study of finding novel DDIs confirms its practical ability, where 7 out of the top 10 candidates are validated in DrugBank.