DERNA Enables Pareto Optimal RNA Design.

Area of Science:

• Computational Biology
• Bioinformatics
• Vaccine Development

Background:

• Designing messenger RNA (mRNA) sequences is critical for vaccine development, with numerous possibilities due to codon degeneracy.
• RNA secondary structure influences thermodynamic stability (minimum free energy, MFE) and mRNA half-life.
• Codon usage bias (codon adaptation index, CAI) affects translation efficiency, with simultaneous optimization of MFE and CAI being a recent focus.

Purpose of the Study:

• To address the challenge of simultaneously optimizing MFE and CAI for RNA sequence design.
• To introduce the Pareto Optimal RNA Design problem and the DEsign RNA (DERNA) algorithm.
• To identify RNA sequences with optimal trade-offs between thermodynamic stability and translation efficiency.

Main Methods:

• Developed the DEsign RNA (DERNA) algorithm using the weighted sum method to enumerate the Pareto front.
• Employed dynamic programming to optimize convex combinations of MFE and CAI objectives.
• Evaluated DERNA on a benchmark dataset, comparing it against CDSfold and LinearDesign.

Main Results:

• DERNA identifies Pareto optimal RNA sequences, offering no single solution better in both MFE and CAI.
• Compared to CDSfold (MFE-only optimization), DERNA achieved identical MFE with superior CAI.
• DERNA's solution quality matched that of LinearDesign, a recent dual-objective optimization approach.

Conclusions:

• The DERNA algorithm effectively balances MFE and CAI for RNA sequence design.
• This approach holds significant potential for optimizing mRNA vaccine design, including for SARS-CoV-2 spike protein targets.