Scholarlink

Background <p>Archaea, the third realm of life, shows unique genetic and metabolic adaptations that enable its survival in harsh environments. Despite its biotechnological and evolutionary importance, limited research has been done on archaeal promoter prediction compared to bacteria and eukaryotes. Current tools have shortcomings such as relying solely on DNA duplex stability feature encoding, high false-positive rates, low precision and lack of publicly accessible archaeal promoter prediction tool.</p> Results <p>To overcome these limitations, we present “iProm-Archaea”, a CNN-based tool for precise archaeal promoter prediction. Our method systematically assesses different feature encoding schemes and finds K-mer (K = 6) as the best representation to capture promoter motifs. For model training, we utilized experimentally validated archaeal promoters from <i>Sulfolobus solfataricus</i>, <i>Haloferax volcanii</i>, and <i>Thermococcus kodakarensis</i>. And <i>T. Kodakarensis KOD1</i> for independent testing. To enhance interpretability, we incorporated explainable artificial intelligence and performed shapley additive explanations to uncover the most influential motifs contributing to the model’s predictions. iProm-Archaea achieves 92% accuracy on training data and 89% on an independent test dataset, outperforming state-of-the-art models. Cross-organism analysis reveals limited generalizability to prokaryotic and eukaryotic promoters, underscoring the distinct regulatory architecture of archaea. To facilitate practical use, we have developed a user-friendly webserver. Moreover, we leverage iProm-Archaea and annotate 5,86,455 archaeal promoters from 478 unannotated archaea genomes.</p> Conclusions <p>This work advances archaeal genomics by offering a robust, domain-specific tool for archaeal promoter detection. The development of a user-friendly webserver enhances practical accessibility for researchers. Our work not only improves prediction performance but also enhances the understanding of archaeal gene regulation to facilitate future biotechnological and evolutionary studies in this underexplored domain of life.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Characterization of archaeal promoters using explainable and web-based CNN model

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年10月21日

查看全文
  • Muhammad Shujaat,
  • Shi-Qing Mao

摘要

Background

Archaea, the third realm of life, shows unique genetic and metabolic adaptations that enable its survival in harsh environments. Despite its biotechnological and evolutionary importance, limited research has been done on archaeal promoter prediction compared to bacteria and eukaryotes. Current tools have shortcomings such as relying solely on DNA duplex stability feature encoding, high false-positive rates, low precision and lack of publicly accessible archaeal promoter prediction tool.

Results

To overcome these limitations, we present “iProm-Archaea”, a CNN-based tool for precise archaeal promoter prediction. Our method systematically assesses different feature encoding schemes and finds K-mer (K = 6) as the best representation to capture promoter motifs. For model training, we utilized experimentally validated archaeal promoters from Sulfolobus solfataricus, Haloferax volcanii, and Thermococcus kodakarensis. And T. Kodakarensis KOD1 for independent testing. To enhance interpretability, we incorporated explainable artificial intelligence and performed shapley additive explanations to uncover the most influential motifs contributing to the model’s predictions. iProm-Archaea achieves 92% accuracy on training data and 89% on an independent test dataset, outperforming state-of-the-art models. Cross-organism analysis reveals limited generalizability to prokaryotic and eukaryotic promoters, underscoring the distinct regulatory architecture of archaea. To facilitate practical use, we have developed a user-friendly webserver. Moreover, we leverage iProm-Archaea and annotate 5,86,455 archaeal promoters from 478 unannotated archaea genomes.

Conclusions

This work advances archaeal genomics by offering a robust, domain-specific tool for archaeal promoter detection. The development of a user-friendly webserver enhances practical accessibility for researchers. Our work not only improves prediction performance but also enhances the understanding of archaeal gene regulation to facilitate future biotechnological and evolutionary studies in this underexplored domain of life.