Immunoglobulin A nephropathy (IgA) is one of the most common primary glomerular diseases and a common predisposing cause of chronic kidney disease. In this paper, using IgA nephropathy as an entry point, we use the New Entity Embedding Representation Algorithm based on Graph Attention Networks (NEEGAT) to solve the problem of the continuous emergence of new entities in the bioinformatics knowledge graph. The algorithm obtains the semantic information in the graph after TransE pre-training, introduces the knowledge graph by means of logical attention, and integrates the inter-section connectivity relations using graph attention to obtain the embedding representation of new entities. The weighted gene co-expression network analysis (WGCNA) method was also selected to identify the modular genes highly associated with IgA nephropathy. The identification and determination of IgA nephropathy was proposed by integrating the pathogenesis of IgA nephropathy under the intestinal-renal axis. Using the designed method to assist the application of the receptor of adenosine A2A in the treatment of IgA nephropathy, the obtained receptor of adenosine A2A was able to promote the increase of the expression level of the receptor α-SMA (up to 53.89% at day 14) and the decrease of the expression level of E-cadherin (up to 40.00% at day 14).
