An interpretable Graph-Regularized Optimal Transport Framework for Diagonal Single-Cell Integrative Analysis.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell omics technologies offer detailed cellular process characterization.
• Limited co-assay sequencing results in unpaired single-cell omics datasets with differing feature dimensions.

Purpose of the Study:

• To present GROTIA (Graph-Regularized Optimal Transport Framework for Diagonal Single-Cell Integrative Analysis), a novel computational method for aligning unpaired multi-omics datasets.

Main Methods:

• GROTIA utilizes optimal transport for global dataset alignment.
• Graph regularization is employed to preserve local cellular relationships.
• Domain-specific feature importance is derived for biological marker identification.

Main Results:

• GROTIA successfully aligns multi-omics datasets without requiring prior correspondence information.
• The method provides interpretable results by highlighting key biological markers.
• Leveraging the transport plan enables post-integration clustering for novel cell subpopulation discovery.

Conclusions:

• GROTIA demonstrates superior performance on simulated and real-world datasets compared to state-of-the-art unsupervised alignment methods.
• The biological significance of identified top features is confirmed.