scGSI: Graph-guided self-supervised integration of paired single-cell multi-omics
- 1School of Computer Science and Engineering, Hunan University of Science and Technology, Xiangtan, Hunan, China
- 2Hunan Provincial Key Laboratory of Intelligent Computing and Language Information Processing, Hunan University of Science and Technology, Xiangtan, Hunan, China
Abstract
Paired single-cell multi-omics technologies provide direct within-cell correspondence across molecular layers and offer a powerful route to dissecting cellular heterogeneity and regulatory relationships. However, effective integration requires more than modality mixing: a useful model must accurately align paired cells while preserving modality-specific topological structure and biologically meaningful variation. Existing methods often struggle with topology mismatch across modalities, underuse cross-modal complementarity within paired cells, or improve alignment at the cost of biological fidelity. To address these challenges, we present scGSI, a graph-guided self-supervised framework for paired single-cell multi-omics integration. scGSI combines heterogeneous graph encoders to preserve modality-specific neighborhood structure, a pull-in projection module to stabilize pre-alignment, and a cross-fusion mechanism with contrastive refinement to exploit complementary signals between paired modalities. Across five paired single-cell multi-omics datasets collected from four platforms, scGSI improves paired cell-state alignment while maintaining a favorable balance between modality mixing and biological variation preservation. The learned embeddings also better support downstream analyses, including cell-type discrimination and developmental trajectory inference, showing that accurate alignment need not erase biologically meaningful structure. Author summary: Single-cell multi-omics technologies now allow us to measure multiple molecular layers (such as gene expression and chromatin accessibility) within the same cell. However, integrating these different data types into a coherent picture remains challenging because each modality has its own statistical properties and biological meaning. We developed scGSI, a computational framework that uses graph-guided self-supervised learning to align paired single-cell multi-omics data while preserving biologically meaningful variation. scGSI works by first maintaining the unique structure of each data type, then progressively aligning them using paired cellular information, and finally refining the integration through contrastive learning. Across five paired datasets from four experimental platforms, scGSI achieves accurate cross-modal alignment while retaining the biological structure needed for downstream analysis. As a result, the integrated embeddings support clearer cell-type identification and more reliable reconstruction of developmental trajectories. This may help researchers interpret complex cell populations and connect molecular measurements with cell identity and development.
Related
- smiMAD: adaptive multi-view graph learning for paired spatial omics integration
- simPIC: flexible simulation of single-cell ATAC-seq paired-insertion counts from individuals to populations.
- AET5: A transcriptome-guided molecular generation framework with contrastive self-supervised learning
- OSCAR: an omics atlas of paired single-cell ATAC-seq and RNA-seq for decoding cell-specific gene regulation
- DR-GEM enables self-supervised machine learning for single-cell embeddings and annotations
BibTeX
@article{chen2026scgsi,
title = {{scGSI: Graph-guided self-supervised integration of paired single-cell multi-omics}},
author = {Xiang Chen and Zihan Yang and Xiaoyu Liu and Zhiyi Xie and Wenlu Guo},
journal = {PLOS Computational Biology},
year = {2026},
month = sep,
doi = {10.1371/journal.pcbi.1014773},
url = {https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1014773},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - scGSI: Graph-guided self-supervised integration of paired single-cell multi-omics
AU - Xiang Chen
AU - Zihan Yang
AU - Xiaoyu Liu
AU - Zhiyi Xie
AU - Wenlu Guo
PY - 2026
DA - 2026/09/08
JO - PLOS Computational Biology
DO - 10.1371/journal.pcbi.1014773
UR - https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1014773
AB - Paired single-cell multi-omics technologies provide direct within-cell correspondence across molecular layers and offer a powerful route to dissecting cellular heterogeneity and regulatory relationships. However, effective integration requires more than modality mixing: a useful model must accurately align paired cells while preserving modality-specific topological structure and biologically meaningful variation. Existing methods often struggle with topology mismatch across modalities, underuse cross-modal complementarity within paired cells, or improve alignment at the cost of biological fidelity. To address these challenges, we present scGSI, a graph-guided self-supervised framework for paired single-cell multi-omics integration. scGSI combines heterogeneous graph encoders to preserve modality-specific neighborhood structure, a pull-in projection module to stabilize pre-alignment, and a cross-fusion mechanism with contrastive refinement to exploit complementary signals between paired modalities. Across five paired single-cell multi-omics datasets collected from four platforms, scGSI improves paired cell-state alignment while maintaining a favorable balance between modality mixing and biological variation preservation. The learned embeddings also better support downstream analyses, including cell-type discrimination and developmental trajectory inference, showing that accurate alignment need not erase biologically meaningful structure. Author summary: Single-cell multi-omics technologies now allow us to measure multiple molecular layers (such as gene expression and chromatin accessibility) within the same cell. However, integrating these different data types into a coherent picture remains challenging because each modality has its own statistical properties and biological meaning. We developed scGSI, a computational framework that uses graph-guided self-supervised learning to align paired single-cell multi-omics data while preserving biologically meaningful variation. scGSI works by first maintaining the unique structure of each data type, then progressively aligning them using paired cellular information, and finally refining the integration through contrastive learning. Across five paired datasets from four experimental platforms, scGSI achieves accurate cross-modal alignment while retaining the biological structure needed for downstream analysis. As a result, the integrated embeddings support clearer cell-type identification and more reliable reconstruction of developmental trajectories. This may help researchers interpret complex cell populations and connect molecular measurements with cell identity and development.
SN - 1553-7358
ER -