TriTan: an efficient triple nonnegative matrix factorization method for integrative analysis of single-cell...

Area of Science:

• Computational Biology
• Genomics
• Systems Biology

Background:

• Single-cell multiomics enables simultaneous profiling of transcriptomes, epigenomes, and proteomes.
• Existing integrative analysis methods are computationally expensive and lack interpretability.
• Challenges exist in applying current methods to large-scale studies for understanding regulatory mechanisms.

Purpose of the Study:

• To develop an efficient and interpretable computational method for single-cell multiomics data integration.
• To facilitate the understanding of gene regulatory networks and cell states.

Main Methods:

• Proposed TriTan (Triple Integrative fast non-negative matrix factorization), an efficient joint factorization method.
• Implemented a highly efficient factorization algorithm to improve computational performance.
• Utilized three matrix factorizations for cell clustering, signature feature identification, and cross-omics association uncovering.

Main Results:

• TriTan demonstrated highly competitive performance compared to state-of-the-art methods on diverse single-cell multiomics datasets.
• The method facilitates the identification of regulatory chromatin domains and prediction of cell-type-specific regulatory networks.
• Downstream analyses highlighted TriTan's capacity to enhance biological discovery and interpretation.

Conclusions:

• TriTan offers an efficient and interpretable solution for single-cell multiomics data analysis.
• The method advances the understanding of gene regulatory networks and cell states.
• TriTan supports large-scale studies and facilitates biological discovery through enhanced interpretation.