Most cancer research tools are built for common cancers with abundant data but they don't work well for rare, aggressive subtypes where data is scarce. Neuroendocrine Prostate Cancer (NEPC) is one such case: it affects around 20,000 people in the US every year, carries a median survival of just 7 months, and is severely underrepresented across every major oncology database.

Standard platforms like Open Targets can tell you what is associated with a disease, but not why- leaving researchers to piece together the underlying biology themselves. This whitepaper shows how Elucidata's Polly Knowledge Graph takes a different approach.

In a real NEPC case study, this approach uncovered a previously unknown gene interaction that revealed a promising druggable target, something standard databases completely missed. The finding was independently validated and presented at AACR 2026.

‍What's Inside the Whitepaper

• Beyond co-occurrence: Finding drug targets in rare cancers with disease-specific knowledge graphs..

• Integrated 20+ public sources, 8M full-text papers, and proprietary data into a single disease-specific knowledge graph with directional relationships and confidence scoring.
• Identified a novel LCOR–EHMT2 interaction that was absent from standard databases and not present in the initial target list through graph-based link prediction.
• Prioritized EHMT2 as a druggable target with active pre-clinical chemistry, whereas standard approaches and Open Targets surfaced only the already-known target EZH2.