cellHarmony: cell-level matching and holistic comparison of single-cell transcriptomes.

Area of Science:

• Genomics
• Computational Biology
• Systems Biology

Background:

• Understanding molecular pathogenesis requires precise analysis of disease-associated cell populations.
• Single-cell genomics is crucial for comparing normal and diseased transcriptional cell states.

Purpose of the Study:

• To develop an integrated solution for unsupervised analysis, classification, and comparison of cell types from diverse single-cell RNA-Seq datasets.
• To identify molecular and cellular origins of complex human diseases.

Main Methods:

• Created cellHarmony, a Python package and AltAnalyze workflow.
• Employed community-clustering and alignment strategy for single-cell transcriptome matching.
• Computed cell-type specific gene expression differences across multiple populations.

Main Results:

• cellHarmony accurately matches single-cell transcriptomes and identifies distinct/shared gene programs.
• The tool identified impacted pathways and regulatory networks for systems-level perturbation analysis.
• Demonstrated improved performance compared to alternative label projection methods.
• Successfully identified cellular origins of malignant states and stratified patients into clinical subtypes.

Conclusions:

• cellHarmony provides a powerful approach for revealing molecular and cellular origins of complex diseases.
• The tool facilitates the identification of disease networks impacting specific cell types and illuminates therapeutic mechanisms.