Label Propagation Prediction of Drug-Drug Interactions Based on Clinical Side Effects

By Ping Zhang, Fei Wang, Jianying Hu, Robert Sorrentino. 2015

233 citations recorded by the source corpus.

Drug-drug interaction (DDI) is an important topic for public health and thus attracts attention from both academia and industry. Here we hypothesize that clinical side effects (SEs) provide a human phenotypic profile and can be translated into the development of computational models for predicting adverse DDIs. We propose an integrative label propagation framework to predict DDIs by integrating SEs extracted from package inserts of prescription drugs, SEs extracted from FDA Adverse Event Reporting System and chemical structures from PubChem. Experimental results based on hold-out validation demonstrated the effectiveness of the proposed algorithm. In addition, the new algorithm also ranked drug information sources based on their contributions to the prediction, thus not only confirming that SEs are important features for DDI prediction but also paving the way for building more reliable DDI prediction models by prioritizing multiple data sources. By applying the proposed algorithm to 1,626 small-molecule drugs which have one or more SE profiles, we obtained 145,068 predicted DDIs. The predicted DDIs will help clinicians to avoid hazardous drug interactions in their prescriptions and will aid pharmaceutical companies to design large-scale clinical trial by assessing potentially hazardous drug combinations. All data sets and predicted DDIs are available at http://astro.temple.edu/~tua87106/ddi.html.

Subfields: Computational Theory and Mathematics, Molecular Biology, Pharmacology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Theory and Mathematics

In Molecular Biology

In Pharmacology

Citations in the corpus

Modeling polypharmacy side effects with graph convolutional networks (2018, methodology): “The methods aggregate similarity measures through clustering or label propagation in order to identify potential drug-drug interactions (Zhang et al., 2015; Ferdousi et al., 2017; Zhang et al., 2017).”
Graph embedding on biomedical networks: methods, applications and evaluations (2019, background): “A considerable amount of efforts has been devoted to developing computational methods to predict potential interactions in various biomedical networks, such as the DDA network (Zhang et al., 2017a), DDI network (Zhang et al., 2015), and PPI network (Wang et al., 2014).”
Mining Electronic Health Records: A Survey (2017, background): “…al. 2014 [Sathyanarayana et al. 2014; White et al. 2013; Iyer et al. 2014; Haerian et al. 2012; Pathak et al. 2013; Carroll et al. 2011; Xu et al. 2011a; Bobo et al. 2014; Zhang et al. 2015; Melton and Hripcsak 2005] Utilizing data mining techniques in certain application areas is very unnatural.”
Predicting potential drug-drug interactions by integrating chemical, biological, phenotypic and network data (2017, methodology): “Zhang [23] As shown in Table 5, our ensemble methods produce better results than other state-of-the-art methods in terms of different metrics.”
Drug Similarity Integration Through Attentive Multi-view Graph Auto-Encoders (2018, methodology): “[Zhang et al. , 2015] proposed an integrative label propagation algorithm to infer clinical side effects from multiple sources with considering high-order similarity.”
Predicting drug–drug interactions through drug structural similarities and interaction networks incorporating pharmacokinetics and pharmacodynamics knowledge (2017, methodology): “Recently, a computational model for predicting DDI was developed through integrated clinical side effect information from the drug labels and FDA adverse event reporting system [18].”
AI in Health: State of the Art, Challenges, and Future Directions (2019, methodology): “[94] developed a label propagation algorithm to predict drug-drug interactions using drug similarity graphs obtained from side-effect profiles in FAERS.”