Enhancing Gene-Disease Associations with Geometric Embedding Methods

Our work "Enhancing Region-Based Geometric Embedding for Gene-Disease Associations" has been accepted and presented in ACM CODS-COMAD 2024, which happened in Bengaluru, India in Jan 2024. In this work, we tried to use the geometrical knowledge graph embedding (KGE) method to represent biomedical entities as n-dimensional convex-shaped objects in vector space inspired by Kulmanov et al. work(link in paper reference) to perform gene-disease association prediction. We achieved encouraging results. The model is based on the principles of EL++ Description Logic and we used HPO (Human Phenotype Ontology) as our KG/dataset. Link of our paper: https://dl.acm.org/doi/10.1145/3632410.3632489