Training mRNA language models for 25 species at a cost of $165: building an AI pipeline from structure prediction to codon optimization
Original title:Training mRNA Language Models Across 25 Species for $165
AI summary
The OpenMed team built an end-to-end AI pipeline covering protein structure prediction, sequence design, and codon optimization. In the codon optimization stage, the CodonRoBERTa-large-v2 model significantly outperformed other architectures with a perplexity of 4.10 and a Spearman CAI correlation of 0.40. The work scaled training to 25 species, training 4 production-grade models in only 55 GPU hours, and established a unique species-conditioning system that enables rapid conversion from a protein concept to synthesis-ready DNA sequences. The full code and experimental results have been open-sourced.
Key Takeaways
- The OpenMed team built an end-to-end AI pipeline covering protein structure prediction,sequence…
- In the codon optimization stage,the CodonRoBERTa-large-v2 model significantly outperformed other…
- 10 and a Spearman CAI correlation of 0.
- The work scaled training to 25 species,training 4 production-grade models in only 55 GPU hours,and…
- The full code and experimental results have been open-sourced.
Quick Facts
- Published
- Updated
- Source
- Hugging Face 社区博客(混合发现)
- Category
- Other
- AI score
- 83
低成本开源生物AI管道,可加速蛋白质工程和药物开发。
Related entities
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: Training mRNA language models for 25 species at a cost of $165: building an AI pipeline from structure prediction to codon optimization. March 31, 2026. https://new.nat6.net/en/news/cmoegbhaj008wslxxhovhk4yr/