Article Details

Abstract

To address the challenges of high computational burden and loss of fine-grained details when processing high-resolution Immunohistochemistry (IHC) images for Protein Subcellular Localization (PSL) prediction, this paper proposes a multi-label prediction model based on a Sparse Self-Attention mechanism, named SSA-PLoc. The model employs a hierarchical Sparse Self-Attention encoding architecture. By setting different sparsity rates, it computes Self-Attention within local windows, which effectively reduces computational complexity and focuses on key subcellular structures. Experimental results on the Vislocas dataset demonstrate that the proposed model achieves a subset accuracy of 61.29% with full-resolution input and attains competitive performance across multiple PSL evaluation metrics, providing a feasible solution for accurate protein subcellular localization analysis from large-scale biological images.

Keywords

protein subcellular localization; immunohistochemistry image; Sparse Self-Attention; multi-label classification

References

1. [1] VAN DYCK C H，SWANSON C J，AISEN P，et al. Lecanemab in Early Alzheimer's Disease [J].New England Journal of Medicine，2023，388（1）：9-21.
2. [2] Han Qing, Zhang Dawei Virtual simulation experiment on subcellular localization observation of plant proteins [J]. Experimental Science and Technology, 2024, 22 (4): 84-89
3. [3] HE K，ZHANG X，REN S，et al. Deep Residual Learning for Image Recognition [C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas：IEEE，2016：770-778.
4. [4] DOSOVITSKIY A. An Image Is Worth 16x16 Words：Transformers for Image Recognition at Scale [J].arXiv:2010.11929 [cs.CV].（2021-06-03）.https://arxiv.org/abs/2010.11929?featured_on=pythonbytes.
5. [5] LIU Z，WANG Z，DU B. Multi-Marginal Contrastive Learning for Multi-Label Subcellular Protein Localization [C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans：IEEE，2022：20626-20635.
6. [6] WEN J W，ZHANG H L，DU P F. VisLocAS：Vision Transformers for Identifying Protein Subcellular Mis-Localization Signatures of Different Cancer Subtypes From Immunohistochemistry Images [J/OL].Computers in Biology and Medicine，2024，174：108392（2024-04-11）.https://doi.org/10.1016/j.compbiomed.2024.108392.
7. [7] SU J，ZHU M，MURTADHA A，et al. ZLPR：A Novel Loss for Multi-Label Classification [J/OL].arXiv:2208.02955 [cs.LG].（2022-08-05）.https://arxiv.org/abs/2208.02955v1.
8. [8] HU J X，YANG Y，XU Y Y，et al. GraphLoc：A Graph Neural Network Model for Predicting Protein Subcellular Localization From Immunohistochemistry Images [J].Bioinformatics，2022，38（21）：4941-4948.
9. [9] HUANG G，LIU Z，VAN DER MAATEN L，et al. Densely Connected Convolutional Networks [C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Honolulu：IEEE，2017：4700-4708.
10. [10] LIU Z，LIN Y，CAO Y，et al. Swin Transformer：Hierarchical Vision Transformer Using Shifted Windows [C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. Montreal：IEEE，2021：10012-10022.
11. [11] TOUVRON H，CORD M，JÉGOU H. DeiT III：Revenge of the ViT [C]//European Conference on Computer Vision. Cham：Springer，2022：516-533.
12. [12] WANG W，XIE E，LI X，et al. PVT v2：Improved Baselines With Pyramid Vision Transformer [J].Computational Visual Media，2022，8（3）：415-424.
13. [13] MEHTA S，RASTEGARI M. MobileViT: Light-Weight, General-Purpose, and Mobile-Friendly Vision Transformer [J/OL].arXiv:2110.02178v1 [cs.CV].（2021-10-05）. https://arxiv.org/pdf/2110.02178v1.
14. [14] GRAHAM B，EL-NOUBY A，TOUVRON H，et al. LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference [C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. Montreal：IEEE，2021：12259-12269.
15. [15] LI Y，HU J，WEN Y，et al. Rethinking Vision Transformers for MobileNet Size and Speed [C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. Paris：IEEE，2023：16889-16900.

About the authors

Qiao Yang (2000-), male, Han ethnicity, from Puyang, Henan Province, is currently pursuing a master's degree. His research focuses on protein subcellular localization prediction based on immunohistochemistry images; Xiao Ruixi (2001-), female, Han ethnicity, from Ji'an, Jiangxi Province, is currently pursuing a master's degree. Her research focuses on protein subcellular localization prediction based on immunofluorescence images; Hong Yonghui (1999-), male, Han ethnicity, from Fuzhou, Jiangxi Province, is currently pursuing a master's degree. His research focuses on predicting human protein subcellular localization based on cross modal transfer learning.