Original title: KGRACDA: A Model Based on Knowledge Graph from Recursion and Attention Aggregation for CircRNA-disease Association Prediction
Authors: Ying Wang,Maoyuan Ma,Yanxin Xie,Qinke Peng,Hongqiang Lyu,Hequan Sun,Laiyi Fu
This article explores the association between circular RNAs (circRNAs) and human diseases. The researchers propose a new approach, called Knowledge Graph-Enhanced Recursive Aggregation for CircRNA-Disease Association (KGRACDA), to predict this association using deep learning. The existing methods for predicting circRNA-disease association have limitations due to the lack of comprehensive datasets. KGRACDA overcomes this limitation by considering the local depth information in addition to the embedded entities and relations. The researchers construct a comprehensive circRNA-disease knowledge graph using a large-scale, multi-source heterogeneous dataset. They then develop a graph neural network with recursive techniques and attention aggregation to predict circRNA-disease associations. Experimental results show that KGRACDA outperforms other methods by explicitly capturing local depth information. To enhance the convenience of circRNA-disease prediction, an interactive web platform called HNRBase v2.0 is also introduced, allowing users to visualize circRNA data, download data, and predict circRNA-disease associations using KGRACDA.
Original article: https://www.biorxiv.org/content/10.1101/2023.12.04.569883v1