Discovery of Disease Relationships via Transcriptomic Signature Analysis Powered by Agentic AI.

Area of Science:

• Genomics
• Computational Biology
• Systems Biology

Background:

• Current disease classification often misses molecular similarities due to varied clinical presentations.
• Understanding molecular connections is crucial for novel therapeutic strategies.

Purpose of the Study:

• To develop a transcriptomics-driven framework for discovering disease relationships.
• To identify molecular commonalities and functional convergence across diverse diseases.
• To explore therapeutic repurposing opportunities and underlying molecular mechanisms.

Main Methods:

• Utilized GenoMAS, an automated agentic AI system, to analyze over 1,300 disease-condition pairs.
• Developed a novel pathway-based similarity framework integrating multi-database enrichment analysis.
• Constructed a disease similarity network based on transcriptomic data.

Main Results:

• Identified robust gene-level overlaps and novel cross-category disease links.
• Revealed shared biological pathways suggesting functional convergence and molecular mechanisms.
• Demonstrated how background conditions modulate transcriptomic similarity and identified therapeutic repurposing candidates.

Conclusions:

• Agentic AI can scale transcriptomic analysis for mechanistic interpretation across complex disease landscapes.
• The framework facilitates discovery of non-obvious disease relationships and therapeutic targets.
• Publicly accessible data enables further research into disease molecular underpinnings.