Scholarlink

<p>Accurate quality assessment is critical for computational prediction and design of RNA three-dimensional (3D) structures, yet it remains a significant challenge. In this work, we introduce RNArank, a deep learning-based approach to both local and global quality assessment of predicted RNA 3D structure models. For a given structure model, RNArank extracts a comprehensive set of multi-modal features and processes them with a Y-shaped residual neural network. This network is trained to predict two intermediate 2D maps, including the inter-nucleotide contact map and the distance deviation map. These maps are then used to estimate the local and global accuracy. Extensive benchmark tests indicate that RNArank consistently outperforms traditional methods and other deep learning-based methods. Moreover, RNArank demonstrates promising performance in identifying high-quality structure models for targets from the recent CASP15 and CASP16 experiments. We anticipate that RNArank will serve as a valuable tool for the RNA biology community, improving the reliability of RNA structure modeling and thereby contributing to a deeper understanding of RNA function.</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

Quality assessment of RNA 3D structure models using deep learning and intermediate 2D maps

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月21日

查看全文
  • Xiaocheng Liu,
  • Wenkai Wang,
  • Zongyang Du,
  • Ziyi Wang,
  • Zhenling Peng,
  • Jianyi Yang

摘要

Accurate quality assessment is critical for computational prediction and design of RNA three-dimensional (3D) structures, yet it remains a significant challenge. In this work, we introduce RNArank, a deep learning-based approach to both local and global quality assessment of predicted RNA 3D structure models. For a given structure model, RNArank extracts a comprehensive set of multi-modal features and processes them with a Y-shaped residual neural network. This network is trained to predict two intermediate 2D maps, including the inter-nucleotide contact map and the distance deviation map. These maps are then used to estimate the local and global accuracy. Extensive benchmark tests indicate that RNArank consistently outperforms traditional methods and other deep learning-based methods. Moreover, RNArank demonstrates promising performance in identifying high-quality structure models for targets from the recent CASP15 and CASP16 experiments. We anticipate that RNArank will serve as a valuable tool for the RNA biology community, improving the reliability of RNA structure modeling and thereby contributing to a deeper understanding of RNA function.