AI Insight
CodonMamba is a new artificial intelligence model designed to predict and generate optimized mRNA coding sequences (CDS). The model, trained on large datasets of biological sequences, outperformed existing methods on 10 out of 12 mRNA prediction benchmarks and introduces a novel "programmable" design framework that allows users to specify codon preferences at the time of sequence generation rather than during training. This enables researchers to tailor mRNA sequences for different host organisms or applications without retraining the entire model, while coordinating optimization across multiple sequence properties simultaneously.
Why it matters
This technology could accelerate the design of therapeutic mRNAs, vaccines, and biotechnology applications by making it faster and easier to optimize genetic sequences for specific purposes. The ability to adjust codon preferences on-demand without retraining represents a significant practical advance for researchers developing mRNA-based treatments or engineering proteins in different organisms.
Understand the Science
⚠️ Preprint – Noch nicht peer-reviewed
Dieser Artikel wurde noch nicht von unabhängigen Experten begutachtet. Die Ergebnisse sind vorläufig und sollten mit Vorsicht interpretiert werden.
Although mRNA codon language models provide a generalizable framework for biological sequence design, effective CDS design requires both a learned sequence design space that captures biological constraints and context-configurable design preferences. Here we present CodonMamba, a codon language model framework for mRNA prediction and programmable CDS design. Pretrained on large-scale coding-sequence corpora, CodonMamba achieved state-of-the-art performance across a comprehensive benchmark of 12 mRNA prediction tasks, ranking first on 10 of 12 tasks. Furthermore, CodonMamba establishes a programmable CDS design framework, transforming codon optimization from a process dependent on preferences embedded during model training into an inference-time steerable generation framework. By introducing user-specified codon usage priors during generation, CodonMamba enables inference-time steering toward host- or application-specific codon preferences without retraining. In design experiments, CodonMamba enabled coordinated optimization over multiple design-relevant sequence properties and demonstrated programmable cross-host CDS retargeting through inference-time prior switching, while preserving most model-derived codon choices. Together, these results establish CodonMamba as a programmable foundation model framework for CDS design, enabling context-specific mRNA sequence engineering and highlighting a promising direction for precise mRNA design using foundation models.
Source: CodonMamba: a foundation model for programmable mRNA coding sequence design