Design of function-regulating RNA via deep learning and AlphaFold 3.

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology
• Synthetic Biology

Background:

• RNAs are crucial regulatory molecules, but their complex structure-function relationships challenge effective RNA design.
• Developing programmable RNAs with predictable functions is essential for various biological applications.

Purpose of the Study:

• To develop a computational framework for enhancing sequence diversity in single-guide RNA (sgRNA) design.
• To improve the efficiency and accuracy of gene editing using computationally designed RNAs.
• To explore the utility of AlphaFold 3 in the one-shot design of CRISPR RNA (crRNA).

Main Methods:

• Integration of deep learning and energy-based computational methods for sgRNA sequence design.
• Utilizing molecular dynamic simulations to assess the stability of DNA-RNA-protein complexes.
• Leveraging AlphaFold 3 confidence metrics to identify functional RNA sequences.

Main Results:

• Achieved high gene editing efficiencies: up to 75% for gene knockouts, 100% for large fragment deletions, and 62.5% for multiplex gene editing.
• Demonstrated that DNA-RNA-protein complex stability is critical for the functionality of designed RNAs.
• Showcased AlphaFold 3's capability to distinguish functional sequences for one-shot crRNA design.

Conclusions:

• The developed computational framework offers an efficient strategy for designing regulatory RNAs with complex interactions.
• AlphaFold 3 shows significant potential in advancing the field of RNA design and engineering.
• This work paves the way for more sophisticated and precise RNA-based therapeutics and research tools.