A major new review article led by researchers from the School of Mathematics at Harbin Institute of Technology has just been published in Quantitative Biology, offering a comprehensive comparison of computational methods for integrating single-cell RNA sequencing (scRNA-seq) and ATAC sequencing (scATAC-seq) data.
Titled “A comparison of integration methods for single‐cell RNA sequencing data and ATAC sequencing data”, this paper provides a timely and in-depth analysis of the rapidly evolving field of single-cell multi-omics integration (Figure 1). As technologies advance to profile multiple molecular layers—such as gene expression and chromatin accessibility—researchers face increasing challenges in harmonizing and interpreting complex datasets.
This study systematically evaluates 16 leading integration methods, including tools for both paired (same-cell) and unpaired (cross-sample) data. The authors assess performance using real-world benchmark datasets from mouse and human cell atlases, focusing on key tasks such as cell clustering, label transfer, and visualization.
Key findings reveal that deep learning-based approaches, such as scAI and DCCA, outperform traditional methods in preserving biological signals and identifying cell types. Meanwhile, Seurat v4 remains a powerful and widely used platform, though it may struggle with datasets containing subtle or rare cell subpopulations. For unpaired data integration, scJoint and scGCN emerged as top performers, offering robust alignment across modalities.
The review also highlights critical challenges in the field, including scalability, data sparsity, and modality-specific noise, and calls for the development of more efficient, interpretable, and biologically aware computational tools to handle atlas-scale datasets.
This open-access article is expected to serve as a key reference for computational biologists, data scientists, and biomedical researchers applying multi-omics integration in areas such as cancer research, immunology, and developmental biology.
DOI: 10.1002/qub2.91