CelLink: integrating single-cell multi-omics data with weak feature linkage and imbalanced cell populations

Area of Science:

• Computational Biology
• Genomics
• Proteomics

Background:

• Single-cell multi-omics technologies offer deep insights into cellular functions.
• Integrating multi-omics data is challenging due to feature linkage and cell population imbalance.

Purpose of the Study:

• To develop a novel method, CelLink, for efficient single-cell multi-omics data integration.
• To address simultaneous challenges of weak feature linkage and imbalanced cell populations.

Main Methods:

• CelLink employs normalization and smoothing for feature profile alignment.
• A multi-phase pipeline using optimal transport algorithm refines cell correspondences.
• It dynamically identifies and excludes unreliable cell matches to prevent imputation errors.

Main Results:

• CelLink demonstrates superior performance in data mixing, cell manifold preservation, and feature imputation accuracy.
• The method significantly outperforms existing approaches on imbalanced cell populations.
• It enables cell subtype annotation, cell mislabeling correction, and spatial transcriptomic analysis.

Conclusions:

• CelLink provides a robust solution for single-cell multi-omics data integration, especially with imbalanced datasets.
• Its capabilities are pivotal for advancing single-cell multi-modal foundation models and spatial cellular biology.