Hugging Face 发布 Carbon-A：跨物种基因预测模型
原标题：Carbon-A: Finding genes in known and unknown genomes
AI 摘要
Hugging Face Bio 发布 Carbon-Annotator（Carbon-A），一个 1.2B 参数模型，可直接从 DNA 序列预测蛋白质编码区域，覆盖哺乳动物、脊椎动物、无脊椎动物、植物、真菌和原生生物，支持 98,304 碱基对上下文窗口并在双链上给出核苷酸分辨率预测。在 42 个基准基因组上，其宏观平均核苷酸 F1 达 0.944，优于 AUGUSTUS、Helixer、Tiberius、ANNEVO、OrionGeno、SegmentNT 和 NTv3 等基线。团队还通过 Iso-Seq 湿实验验证了部分与 RefSeq 参考注释不一致的新预测基因，表明模型可能发现遗漏注释而非误报。
正文节选
You can explore the Carbon Annotation Database, the model, the training data, and the technical report in HuggingFaceBio’s Carbon Annotation Database Collection. We’re seeing signs that the major AI labs are gearing up to tackle biology as the next frontier after cyber and math. We think 2027 will be the year for bio. It has the potential to improve our lives far beyond what we’ve seen from AI in other fields so far. But we’ve also seen how fragile the ecosystem can become when researchers depen