This research demonstrates how large language models can independently rediscover the central dogma of molecular biology through multilingual transfer learning approaches.

• LLMs trained on biological sequences can identify patterns in DNA, RNA, and protein relationships
• Models successfully capture the genetic code translation rules between nucleotide triplets and amino acids
• Cross-sequence prediction capabilities reveal fundamental biological principles without explicit programming
• Findings suggest LLMs can serve as powerful tools for both confirming established biological laws and potentially discovering new ones

This work highlights the emerging role of AI in validating core biological principles and offers promising avenues for future discoveries in molecular biology and genetics.

Find Central Dogma Again: Leveraging Multilingual Transfer in Large Language Models