RNA Knowledge Graph Analysis via Embedding Methods

Recent advances in RNA technologies opened the avenue to the design of novel vaccines as witnessed by the success of the COVID-19 vaccine and also by new ongoing vaccines for cancer. New drugs based on non-coding RNA can also be developed at lower costs considering the relatively simple structure of these molecules with respect to classical recombinant protein technologies. We recently developed RNA-KG, a biomedical Knowledge Graph focused on RNA, collecting information from more than 50 public databases and bio-medical ontologies to support the study of RNA and the design of novel RNA-based drugs. In this work we show that, by applying inductive machine learning methods on top of embedded node and edges obtained by applying classical Graph Representation Learning methods, we can accurately predict the entities and the relationships between entities included in RNA-KG. Our results open the way to the analysis and the discovery of novel relationships between RNAs and other bio-molecules and medical concepts represented in RNA-KG.