Scholarlink

Background <p>N6-methyladenosine (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(m^{6}A\)</EquationSource> </InlineEquation>) plays a crucial role in enriching RNA functional and genetic information. While deep learning has advanced <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(m^{6}A\)</EquationSource> </InlineEquation> site identification, current state-of-the-art methods, particularly those based on large-scale Pre-trained Language Models (PLMs), often suffer from high computational complexity and over-parameterization, limiting their scalability for genome-wide analysis.</p> Results <p>In this work, we propose MKE-ResNet, an ultra-lightweight end-to-end framework designed to balance predictive performance with extreme computational efficiency. MKE-ResNet integrates a deep residual network with an innovative Multi-Kernel Efficient Channel Attention (MKE-ECA) module to adaptively capture multi-scale sequence patterns. Extensive experiments on 22 datasets demonstrate that MKE-ResNet achieves competitive performance against complex SOTA models (including MST-m6A). Notably, our model exhibits exceptional stability against sequence noise perturbation and demonstrates superior generalization capability on independent test sets (leading in 8 out of 11 cases in terms of AUROC). Furthermore, interpretability analysis reveals that MKE-ResNet moves beyond the static central motif to capture dynamic sequence patterns resembling RBP binding motifs (e.g., SRSF1 and PUM2). The source code and datasets are available at <a href="https://github.com/Gxttk/MKE-Resnet.">https://github.com/Gxttk/MKE-Resnet.</a></p> Conclusions <p>MKE-ResNet provides a biologically interpretable and computationally efficient solution for <InlineEquation ID="IEq3"> <EquationSource Format="TEX">\(m^{6}A\)</EquationSource> </InlineEquation> site identification, offering a pragmatic tool for large-scale epitranscriptome screening.</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

Mke-resnet: a lightweight and interpretable deep learning framework for efficient RNA m6A site identification

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月19日

查看全文
  • Xiao Gao,
  • Jianhua Jia,
  • Cong Hui,
  • Yang Lin

摘要

Background

N6-methyladenosine ( \(m^{6}A\) ) plays a crucial role in enriching RNA functional and genetic information. While deep learning has advanced \(m^{6}A\) site identification, current state-of-the-art methods, particularly those based on large-scale Pre-trained Language Models (PLMs), often suffer from high computational complexity and over-parameterization, limiting their scalability for genome-wide analysis.

Results

In this work, we propose MKE-ResNet, an ultra-lightweight end-to-end framework designed to balance predictive performance with extreme computational efficiency. MKE-ResNet integrates a deep residual network with an innovative Multi-Kernel Efficient Channel Attention (MKE-ECA) module to adaptively capture multi-scale sequence patterns. Extensive experiments on 22 datasets demonstrate that MKE-ResNet achieves competitive performance against complex SOTA models (including MST-m6A). Notably, our model exhibits exceptional stability against sequence noise perturbation and demonstrates superior generalization capability on independent test sets (leading in 8 out of 11 cases in terms of AUROC). Furthermore, interpretability analysis reveals that MKE-ResNet moves beyond the static central motif to capture dynamic sequence patterns resembling RBP binding motifs (e.g., SRSF1 and PUM2). The source code and datasets are available at https://github.com/Gxttk/MKE-Resnet.

Conclusions

MKE-ResNet provides a biologically interpretable and computationally efficient solution for \(m^{6}A\) site identification, offering a pragmatic tool for large-scale epitranscriptome screening.