KG-PREDICT

KG-Predict is a knowledge graph computational framework for drug repurposing that integrates multiple types of genotypic and phenotypic data. The framework constructs GP-KG (Genotype-Phenotype Knowledge Graph), containing 1,246,726 associations between 61,146 biomedical entities from various databases. KG-Predict uses graph embedding methods to learn low-dimensional representations of entities and relations, enabling inference of new drug-disease interactions. The system has been validated for identifying repositioned candidate drugs, particularly for Alzheimer's disease, achieving high performance metrics (AUROC = 0.981, AUPR = 0.409) and successfully prioritizing FDA-approved and clinical trial anti-AD drugs.

Graph at a glance

Node and edge counts are not reported in the registry for this graph.

Products & downloads