基于多层语义与拓扑融合的异质图方法提升药物-靶标相互作用预测性能

打开文本图片集

Abstract：ObjectiveTodevelopaheterogeneous graphpredictionmethodbasedonthefusionof multi-layer semanticsand topologicalinformatonfoadressingthechallengesinug-aretinteractionpdictionncudinginsuientoelingof high-ordersemanticdependencies，lackofadaptivefusionofsmanticpaths，andover-smothingofnodefeatures.Methods Aheterogeneousgraphnetworkwithmultipletypesofentitiessuchasdrugs，proteins，sideefects，anddiseaseswas constructed，andgraph embedding techniqueswereused toobtainlow-dimensional featurerepresentations.Anadaptive metapathsearchmodulewasintroduced toautomaticalldiscoversemanticpathcombinationsforguidingthepropagationof high-order semantic information.Asemanticaggegation mechanism integratingmulti-head atention was designed to automatically learn theimportanceof each semanticpath basedoncontextual informationandachieve diferentiated aggregation and dynamic fusion among paths.A structure-aware gated graph convolutional module was then incorporated to regulate the feature propagation intensityfor suppressing redundantinformation andredcuing over-smoothing.Finalythe potentialinteractionsbetweendrugsandtargets werepredictedthroughaninner productoperation.Results Comparedwith existing drug-target interactionpredictionmethods，theproposed methodachievedanaverage improvement of 3.4% and 2.4% 3.0% and 3.8% in terms of the area under the receiver operating characteristic curve （AUC） and thearea under the precision-recallcurve （AUPRC）onpublicdatasets，respectively.ConclusionThedrug-target interactionpredictionmethod developedinthisstudycanefectivelyextractcomplexhighordersemanticandtopologicalinformationfromheterogeneous biologicalnetworks，therebyimprovingtheaccuracyandstabilityofdrug-targetinteractionprediction.Thismethodprovides technicalsupportand theoretical foundation for precise drug target discoveryandtargeted treatment ofcomplex diseases.

Keywords：drug-targetinteraction;heterogeneousnetworks;gatedmechanism;multi-eadatentionmechansm;graph convolutional networks

药物与靶标相互作用（DTI）的预测研究可促进人们对药物耐药性、副作用的理解，在医学和药物研究领域具有重要意义。（剩余18672字）