Omni-DNA introduces a cross-modal, multi-task genomic foundation model that eliminates the need for separate fine-tuning across different genomic applications.

• Creates a single versatile model capable of handling diverse genomic tasks
• Supports flexible output formats beyond the constraints of existing genomic models
• Demonstrates how transformer-based auto-regressive approaches can be optimized for DNA sequence analysis
• Offers significant efficiency gains as genomic model sizes continue to grow

Why it matters: This approach mirrors the evolution seen in Large Language Models, bringing the same versatility and efficiency to genomic analysis. This breakthrough could accelerate research in personalized medicine, genetic disease studies, and functional genomics.

Omni-DNA: A Unified Genomic Foundation Model for Cross-Modal and Multi-Task Learning