UINMF performs mosaic integration of single-cell multi-omic datasets using nonnegative matrix factorization.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell genomic technologies offer powerful insights into cellular heterogeneity.
• Integrating diverse single-cell datasets is challenging due to varying features (genes/cells) across experiments.
• Existing methods often require shared features, limiting comprehensive data integration.

Purpose of the Study:

• To develop a novel computational method for mosaic integration of single-cell datasets.
• To enable the inclusion of both shared and unshared features across datasets.
• To improve the accuracy and scope of single-cell data integration.

Main Methods:

• Developed a nonnegative matrix factorization algorithm (UINMF).
• Incorporated an additional metagene matrix to leverage unshared features.
• Applied the algorithm to diverse single-cell data types including RNA-seq, spatial transcriptomics, and cross-species data.

Main Results:

• The UINMF algorithm effectively integrates single-cell datasets with unshared features.
• Demonstrated significant improvements in data integration across multiple single-cell technologies.
• Successfully incorporated the UINMF algorithm into the open-source LIGER R package.

Conclusions:

• The UINMF algorithm overcomes limitations of previous methods by utilizing unshared features.
• This approach enhances the power of single-cell genomics for defining molecular cell types.
• The LIGER R package provides accessible implementation for researchers.