Designing RNA sequences that encode a given protein while also folding into a desired structure is a combinatorial challenge. Codon optimization for mRNA therapeutics and vaccine design requires navigating a vast space of synonymous codons, and existing computational methods typically rely on simplified objectives that may not capture detailed secondary structure thermodynamics.
The authors introduce a direct and efficient algorithm that samples sequences from a Boltzmann distribution defined by both the codon sequence and a fully detailed secondary-structure free energy model. Building on a recently developed tensor-based formulation of RNA thermodynamics, the method also enables exact computation of statistical quantities such as free energies, base-pairing probabilities, and base and codon marginals.
According to the paper, this is the first time global sequence design has been accomplished with respect to such a highly accurate free energy model. The algorithms are designed to leverage any available CPU and GPU resources in parallel, offering substantial computational speedups for practical applications.