Collected molecules will appear here. Add from search or explore.
Knowledge graph embedding framework for drug repurposing and discovery specifically targeted at rare diseases, using a trimodal approach to integrate diverse biomedical data.
Defensibility
stars
3
RDKG-115 is a static research repository representing a specific paper's methodology from approximately late 2020. With only 3 stars and zero forks over nearly four years, the project has no community traction, developer adoption, or maintenance velocity. Its defensibility is near zero as it is a standard academic code dump that implements known KGE patterns (like TransE or RotatE variants) applied to a specific niche dataset. While the 'trimodal' approach was a useful research contribution at the time, the field of drug discovery AI has moved significantly toward more advanced Graph Neural Networks (GNNs), geometric deep learning, and foundation models (e.g., BioMedLM, Med-PaLM) that can perform cross-domain reasoning more effectively than static embeddings. Competitive tools from well-funded biotech startups (like Recursion or Insilico Medicine) or specialized discovery platforms (like BenchSci or Causaly) represent a significant hurdle for this type of standalone academic project. The displacement risk is high because the core capability—predicting drug-disease links—is now being absorbed into much larger, more integrated biomedical LLMs and graph-informed foundation models.
TECH STACK
INTEGRATION
reference_implementation
READINESS