ST-deconv:利用自编码和对比学习的空间转录组数据的精确反褶积方法。

IF 2.8 Q1 GENETICS & HEREDITY
NAR Genomics and Bioinformatics Pub Date : 2025-08-27 eCollection Date: 2025-09-01 DOI:10.1093/nargab/lqaf109
Shurui Dai, Jiawei Li, Zhiliang Xia, Jingfeng Ou, Yan Guo, Limin Jiang, Jijun Tang
{"title":"ST-deconv:利用自编码和对比学习的空间转录组数据的精确反褶积方法。","authors":"Shurui Dai, Jiawei Li, Zhiliang Xia, Jingfeng Ou, Yan Guo, Limin Jiang, Jijun Tang","doi":"10.1093/nargab/lqaf109","DOIUrl":null,"url":null,"abstract":"<p><p>Single-cell RNA sequencing (scRNA-seq) has significantly deepened our understanding of cellular heterogeneity and cell type interactions, providing insights into how cell populations adapt to environmental variability. However, its lack of spatial context limits intercellular analysis. Similarly, existing spatial transcriptomics (ST) data often lack single-cell resolution, restricting cellular mapping. To address these limitations, we introduce ST-deconv, a deep learning-based deconvolution model that integrates spatial information. ST-deconv leverages contrastive learning to enhance the spatial representation of adjacent spots, improving spatial relationship inference. It also employs domain-adversarial networks to improve generalization and deconvolution across diverse datasets. Moreover, ST-deconv can generate large-scale, high-resolution spatial transcriptomic data with cell type labels from single-cell input, facilitating the learning of spatial cell type composition. In benchmarking experiments, ST-deconv outperforms traditional methods, reducing the root mean square error (RMSE) by 13% to 60%, with an RMSE as low as 0.03 for high spatial correlation datasets and 0.07 for low spatial correlation datasets across different transcriptomic contexts. Reconstructing real tissue structure, a purity of 0.68 on mouse olfactory bulb (MOB) and a cell type correlation of 0.76 on human pancreatic ductal adenocarcinoma (PDAC) were achieved. These advancements make ST-deconv a powerful tool for enhancing spatial transcriptomics and downstream analyses of intercellular interactions.</p>","PeriodicalId":33994,"journal":{"name":"NAR Genomics and Bioinformatics","volume":"7 3","pages":"lqaf109"},"PeriodicalIF":2.8000,"publicationDate":"2025-08-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12390763/pdf/","citationCount":"0","resultStr":"{\"title\":\"ST-deconv: an accurate deconvolution approach for spatial transcriptome data utilizing self-encoding and contrastive learning.\",\"authors\":\"Shurui Dai, Jiawei Li, Zhiliang Xia, Jingfeng Ou, Yan Guo, Limin Jiang, Jijun Tang\",\"doi\":\"10.1093/nargab/lqaf109\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><p>Single-cell RNA sequencing (scRNA-seq) has significantly deepened our understanding of cellular heterogeneity and cell type interactions, providing insights into how cell populations adapt to environmental variability. However, its lack of spatial context limits intercellular analysis. Similarly, existing spatial transcriptomics (ST) data often lack single-cell resolution, restricting cellular mapping. To address these limitations, we introduce ST-deconv, a deep learning-based deconvolution model that integrates spatial information. ST-deconv leverages contrastive learning to enhance the spatial representation of adjacent spots, improving spatial relationship inference. It also employs domain-adversarial networks to improve generalization and deconvolution across diverse datasets. Moreover, ST-deconv can generate large-scale, high-resolution spatial transcriptomic data with cell type labels from single-cell input, facilitating the learning of spatial cell type composition. In benchmarking experiments, ST-deconv outperforms traditional methods, reducing the root mean square error (RMSE) by 13% to 60%, with an RMSE as low as 0.03 for high spatial correlation datasets and 0.07 for low spatial correlation datasets across different transcriptomic contexts. Reconstructing real tissue structure, a purity of 0.68 on mouse olfactory bulb (MOB) and a cell type correlation of 0.76 on human pancreatic ductal adenocarcinoma (PDAC) were achieved. These advancements make ST-deconv a powerful tool for enhancing spatial transcriptomics and downstream analyses of intercellular interactions.</p>\",\"PeriodicalId\":33994,\"journal\":{\"name\":\"NAR Genomics and Bioinformatics\",\"volume\":\"7 3\",\"pages\":\"lqaf109\"},\"PeriodicalIF\":2.8000,\"publicationDate\":\"2025-08-27\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12390763/pdf/\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"NAR Genomics and Bioinformatics\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1093/nargab/lqaf109\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"2025/9/1 0:00:00\",\"PubModel\":\"eCollection\",\"JCR\":\"Q1\",\"JCRName\":\"GENETICS & HEREDITY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"NAR Genomics and Bioinformatics","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1093/nargab/lqaf109","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2025/9/1 0:00:00","PubModel":"eCollection","JCR":"Q1","JCRName":"GENETICS & HEREDITY","Score":null,"Total":0}
引用次数: 0

摘要

单细胞RNA测序(scRNA-seq)极大地加深了我们对细胞异质性和细胞类型相互作用的理解,为细胞群体如何适应环境可变性提供了见解。然而,它缺乏空间背景限制了细胞间分析。同样,现有的空间转录组学(ST)数据往往缺乏单细胞分辨率,限制了细胞定位。为了解决这些限制,我们引入了ST-deconv,这是一种基于深度学习的反卷积模型,它集成了空间信息。ST-deconv利用对比学习来增强相邻点的空间表征,提高空间关系推断。它还采用域对抗网络来改进不同数据集的泛化和反卷积。此外,ST-deconv可以从单细胞输入中生成大规模、高分辨率的带有细胞类型标记的空间转录组数据,便于对空间细胞类型组成的学习。在基准测试实验中,ST-deconv优于传统方法,将均方根误差(RMSE)降低了13%至60%,在不同转录组背景下,高空间相关性数据集的RMSE低至0.03,低空间相关性数据集的RMSE低至0.07。重建真实组织结构,小鼠嗅球(MOB)的纯度为0.68,人胰腺导管腺癌(PDAC)的细胞类型相关性为0.76。这些进展使ST-deconv成为增强空间转录组学和细胞间相互作用下游分析的有力工具。
本文章由计算机程序翻译,如有差异,请以英文原文为准。

ST-deconv: an accurate deconvolution approach for spatial transcriptome data utilizing self-encoding and contrastive learning.

ST-deconv: an accurate deconvolution approach for spatial transcriptome data utilizing self-encoding and contrastive learning.

ST-deconv: an accurate deconvolution approach for spatial transcriptome data utilizing self-encoding and contrastive learning.

ST-deconv: an accurate deconvolution approach for spatial transcriptome data utilizing self-encoding and contrastive learning.

Single-cell RNA sequencing (scRNA-seq) has significantly deepened our understanding of cellular heterogeneity and cell type interactions, providing insights into how cell populations adapt to environmental variability. However, its lack of spatial context limits intercellular analysis. Similarly, existing spatial transcriptomics (ST) data often lack single-cell resolution, restricting cellular mapping. To address these limitations, we introduce ST-deconv, a deep learning-based deconvolution model that integrates spatial information. ST-deconv leverages contrastive learning to enhance the spatial representation of adjacent spots, improving spatial relationship inference. It also employs domain-adversarial networks to improve generalization and deconvolution across diverse datasets. Moreover, ST-deconv can generate large-scale, high-resolution spatial transcriptomic data with cell type labels from single-cell input, facilitating the learning of spatial cell type composition. In benchmarking experiments, ST-deconv outperforms traditional methods, reducing the root mean square error (RMSE) by 13% to 60%, with an RMSE as low as 0.03 for high spatial correlation datasets and 0.07 for low spatial correlation datasets across different transcriptomic contexts. Reconstructing real tissue structure, a purity of 0.68 on mouse olfactory bulb (MOB) and a cell type correlation of 0.76 on human pancreatic ductal adenocarcinoma (PDAC) were achieved. These advancements make ST-deconv a powerful tool for enhancing spatial transcriptomics and downstream analyses of intercellular interactions.

求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
CiteScore
8.00
自引率
2.20%
发文量
95
审稿时长
15 weeks
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信