EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

GLAlign integrates single-cell multi-omics data via global manifold and local marker gene alignment

Teer Ba iD, Mengfan Hu, Shuo Liu, Lirong Zhang iD, Lu Chai, Zihan Li, Caixia Gao, Yong Wang iD

Briefings in Bioinformatics · 02.11.2026 · 27 (6)

DOI: 10.1093/bib/bbag633 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Briefings in Bioinformatics, online erschienen am 13.11.2026. Verfasst von Teer Ba und 7 weiteren Autor:innen. Beteiligte Einrichtung: School of Physical Science and Technology, Inner Mongolia University, No. 235 West College Road, Saihan District, Hohhot 010021, China. Hauptthemen laut MEDLINE: Single-Cell Analysis, Software, Computational Biology. Die Arbeit nennt 63 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0). Gefördert von: National Key Research and Development Program of China, Shenzhen Medical Research Fund, Strategic Priority Research Program of the Chinese Academy of Sciences.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Integrating heterogeneous single-cell multi-omics datasets remains challenging because different modalities are measured in distinct feature spaces and lack cell level pairing. Existing manifold alignment or feature conversion framework are often affected by weak cross-modality associations, geometric ambiguity, and overcorrection. Here we present GLAlign (Global manifold and Local marker gene Alignment), an optimization framework that jointly leverages global manifold structure and local marker gene information to integrate single-cell multi-omics data. GLAlign learns a shared latent space through a global objective comprising maximum mean discrepancy, distortion, and penalty terms, together with a local contrastive triplet loss guided by cross-modality marker gene overlap between independently inferred clusters. We demonstrate GLAlign’s methodological contribution in extending the manifold alignment by biologically informed prior to ensure geometrically plausible and biologically correct alignments without requiring paired cells or manual cell type annotations. We evaluated GLAlign on simulated data and real datasets spanning PBMC scRNA-seq/scATAC-seq, mouse brain SNARE-seq, single-donor and multi-batch CITE-seq, and NCI-N87 scDNA-seq/scRNA-seq integration. Across the evaluated datasets, GLAlign achieved strong modality mixing and biological conservation, remained stable under moderate cell type imbalance and variations in clustering and marker information, and supported cross-modality label transfer and paired cell recovery. These results demonstrate that biologically structured coupling of global and local alignment provides an interpretable framework for heterogeneous single-cell multi-omics integration. GLAlign is freely available at https://github.com/Bateer-Bio/GLAlign.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Single-Cell AnalysisSoftwareComputational BiologyMultiomicsAnimalsMiceAlgorithmsHumansGenetic MarkersSingle-Cell Gene Expression Analysis

Stichwörter der Autor:innen

contrastive triplet lossglobal manifold alignmentlocal marker gene alignmentmaximum mean discrepancysingle-cell multi-omics integration

Substanzen

Genetic Markers

Autor:innen und Institutionen

  • School of Physical Science and Technology, Inner Mongolia University, No. 235 West College Road, Saihan District, Hohhot 010021, China.
  • School of Mathematical Sciences, Inner Mongolia University, No. 235 West College Road, Saihan District, Hohhot 010021, China.
  • State Key Laboratory of Mathematical Sciences, Academy of Mathematics and Systems Science, Chinese Academy of Sciences, No. 55 Zhongguancun East Road, Haidian District, Beijing 100190, China.
  • School of Mathematics,University of Chinese Academy of Sciences, Chinese Academy of Sciences, No. 1 Yanqihu East Rd, Huairou District, Beijing 100049, China.
  • Key Laboratory of Systems Biology, Hangzhou Institute for Advanced Study, University of Chinese Academy of Sciences, Chinese Academy of Sciences, No. 1, Xiangshan Zhinong, Xihu District, Hangzhou 330106, China.

ORCID-Profile: Teer Ba, Lirong Zhang, Yong Wang

Förderung

  • National Key Research and Development Program of China
  • Shenzhen Medical Research Fund
  • Strategic Priority Research Program of the Chinese Academy of Sciences
  • CAS Project for Young Scientists in Basic Research
  • National Natural Science Foundation of China
  • National Natural Science Foundation of China, Regional Science Foundation Project

Zitieren

APA

Ba, T., Hu, M., Liu, S., Zhang, L., Chai, L., Li, Z., Gao, C., & Wang, Y. (2026). GLAlign integrates single-cell multi-omics data via global manifold and local marker gene alignment. Briefings in Bioinformatics, 27(6). https://doi.org/10.1093/bib/bbag633

Vancouver

Ba T, Hu M, Liu S, Zhang L, Chai L, Li Z, et al. GLAlign integrates single-cell multi-omics data via global manifold and local marker gene alignment. Briefings in Bioinformatics. 2026;27(6). doi:10.1093/bib/bbag633

BibTeX
@article{Ba2026,
  author = {Ba, Teer and Hu, Mengfan and Liu, Shuo and Zhang, Lirong and Chai, Lu and Li, Zihan and Gao, Caixia and Wang, Yong},
  title = {GLAlign integrates single-cell multi-omics data via global manifold and local marker gene alignment},
  journal = {Briefings in Bioinformatics},
  year = {2026},
  volume = {27},
  number = {6},
  doi = {10.1093/bib/bbag633}
}
Kostenlos starten →