Serokell AI experts developed graph neural network models to predict drug-disease interactions using data from Elsevier. The project involved two main tasks: predicting interactions between drugs and diseases using heterogeneous graph structures with 10 node types and 158 edge types, and biological sequence embedding to
•10m read time• From serokell.io
Table of contents
Drug-disease interaction prediction and biological sequence embeddingWhat is graph machine learning?Graph neural networksMessage passingGraph convolutionsSimpleConv and GraphConvHow are GNNs trained?Graph typesData available in the Elsevier projectNavigating challenges: our progress and future plansSort: