Fine scale structural information substantially improves multivariate regression model for mRNA in-vial degradation...

Area of Science:

• Biochemistry
• Molecular Biology
• Bioinformatics

Background:

• The efficacy of COVID-19 mRNA vaccines highlights the need for stable mRNA molecules.
• Understanding sequence metrics that govern mRNA stability in solution is essential for vaccine development and storage.
• Current methods for predicting mRNA stability often overlook local structural variations crucial for degradation resistance.

Purpose of the Study:

• To identify key sequence metrics that influence mRNA stability in solution.
• To develop an improved model for predicting mRNA stability by incorporating local structural features.
• To enhance the design of mRNA molecules for improved in-solution stability.

Main Methods:

• Analysis of RNA secondary structure and its correlation with degradation.
• Utilizing base-pairing probability (log odds) as a fine-scale metric for structural analysis.
• Developing a regression model (STRAND) combining global and local sequence features.

Main Results:

• Local structural features, particularly base-pairing probability, offer orthogonal insights beyond global metrics like minimum free energy.
• The STRAND model, integrating four key features, significantly outperforms existing methods in predicting mRNA stability.
• STRAND provides a compact and interpretable framework for stability-focused mRNA design.

Conclusions:

• Base-pairing probability is a critical, fine-scale metric for predicting mRNA stability.
• The STRAND model offers a practical and effective approach for designing more stable mRNA molecules.
• Enhanced mRNA stability is achievable through a deeper understanding of local structural determinants.