RNA design via structure-aware multifrontier ensemble optimization.

Area of Science:

• Computational biology
• Bioinformatics
• Molecular biology

Background:

• RNA design, the inverse problem of RNA folding, seeks sequences for desired structures.
• Existing algorithms often yield low ensemble stability, especially for long RNA sequences.
• Current methods typically find few sequences meeting Minimum Free Energy (MFE) criteria, limiting applications.

Purpose of the Study:

• To introduce SAMFEO, an innovative optimization paradigm for RNA sequence design.
• To address limitations of existing RNA design algorithms, including low ensemble stability and limited sequence generation.
• To develop a method capable of designing a large number of RNA sequences, particularly for complex structures.

Main Methods:

• SAMFEO employs an iterative search strategy optimizing ensemble objectives like equilibrium probability and ensemble defect.
• The algorithm integrates structure-level and ensemble-level information throughout the search process: initialization, sampling, mutation, and updating.
• This approach is designed to be less complex than existing methods while maximizing sequence design output.

Main Results:

• SAMFEO is the first algorithm to design thousands of RNA sequences for the Eterna100 benchmark puzzles.
• The algorithm solves more Eterna100 puzzles than other general optimization-based methods.
• SAMFEO demonstrates superior performance in designing long RNA sequences for structures from the 16S Ribosomal RNA database.

Conclusions:

• SAMFEO offers a novel and effective approach to RNA sequence design, overcoming key limitations of prior methods.
• The algorithm's ability to generate numerous high-quality RNA sequences, including long ones, expands the possibilities for RNA-based applications.
• SAMFEO represents a significant advancement in computational RNA design, providing a powerful tool for researchers.