Single-round evolution of RNA aptamers with GRAPE-LM

Area of Science:

• Biochemistry
• Molecular Biology
• Artificial Intelligence

Background:

• Directed evolution is crucial for biomolecule development, but RNA evolution lags behind protein evolution.
• Traditional RNA aptamer discovery involves laborious, multi-round screening processes.
• Advancements in language models have accelerated protein evolution, highlighting a need for similar RNA tools.

Purpose of the Study:

• To introduce GRAPE-LM, a novel generative AI framework for one-round RNA aptamer evolution.
• To demonstrate GRAPE-LM's capability in rapidly discovering high-affinity RNA aptamers.
• To overcome the limitations of traditional multi-round screening methods in aptamer selection.

Main Methods:

• GRAPE-LM integrates a transformer-based conditional autoencoder with nucleic acid language models.
• The framework utilizes CRISPR-Cas-based aptamer screening data from intracellular environments for guidance.
• The system is designed for efficient, one-round evolution of RNA aptamers.

Main Results:

• GRAPE-LM successfully generated RNA aptamers for three distinct targets: CD3ε, SARS-CoV-2 RBD, and c-Myc.
• The AI-generated aptamers outperformed those obtained through multiple rounds of human selection and optimization.
• The framework demonstrated efficacy even with limited screening data (single round).

Conclusions:

• GRAPE-LM offers a significant advancement in RNA aptamer evolution, enabling rapid, one-round discovery.
• The AI framework provides a powerful alternative to traditional, time-consuming aptamer selection methods.
• GRAPE-LM's success with challenging targets like intracellular disordered proteins broadens its applicability.