JSNMFuP: a unsupervised method for the integrative analysis of single-cell multi-omics data based on non-negative
Bai Zhang1, Mengdi Nan1, Liugen Wang2
1School of Science, Jiangnan University, Wuxi, Jiangsu, China.
BMC Genomics
|March 21, 2025
Summary
This study introduces JSNMFuP, a new method for integrating single-cell multi-omics data. It effectively reveals cellular heterogeneity and improves cell clustering performance for biological insights.
Area of Science:
- Computational Biology
- Genomics
- Bioinformatics
Background:
- Single-cell multi-omics data offers deep insights into cellular phenotypes.
- Integrating diverse omics data is crucial for understanding cellular heterogeneity.
- Challenges exist in integrating single-cell data due to differing modalities and noise.
Purpose of the Study:
- To develop an unsupervised method for integrating single-cell multi-omics data.
- To enhance the analysis of cellular heterogeneity.
- To improve cell clustering and downstream biological interpretation.
Main Methods:
- Proposed JSNMFuP, an unsupervised integration method based on non-negative matrix factorization.
- Integrated omics information via latent variables using a consensus graph.
- Incorporated high-dimensional geometric structure and cross-modal feature links using regularization.
Main Results:
- JSNMFuP demonstrated superior performance in cell clustering on real datasets.
- The method effectively captures high-dimensional geometric structure.
- Identified interpretable factors for analyzing cell heterogeneity.
Conclusions:
- JSNMFuP is an effective method for single-cell multi-omics data integration.
- Facilitates data visualization, clustering, marker characterization, and gene ontology enrichment.
- Provides valuable biological insights for downstream analysis of cell heterogeneity.
