Joint analysis of heterogeneous single-cell RNA-seq dataset collections

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) generates high-dimensional data from individual cells.
• Analyzing scRNA-seq data across multiple samples (individuals, conditions, tissues) presents challenges in identifying consistent biological features.
• Existing methods often struggle with integrating and comparing cell populations across diverse datasets.

Purpose of the Study:

• To develop a robust computational approach for identifying recurrent cell subpopulations in large, heterogeneous scRNA-seq collections.
• To enable the construction of comprehensive single-cell atlases by integrating data from multiple sources.
• To facilitate the discovery of conserved cell types and states across different biological contexts.

Main Methods:

• Developed Conos, a novel algorithm for scRNA-seq data integration.
• Constructed a global graph by utilizing multiple plausible inter-sample mappings to connect all cells.
• Employed graph-based methods for identifying cell clusters and propagating information across datasets.

Main Results:

• Conos successfully identifies recurrent cell clusters present in multiple datasets.
• The global graph approach effectively integrates heterogeneous scRNA-seq collections.
• Information propagation within the graph allows for enhanced characterization of cell states across samples.

Conclusions:

• Conos provides a powerful framework for analyzing large-scale, multi-sample scRNA-seq data.
• This approach facilitates the discovery of conserved cell subpopulations crucial for building single-cell atlases.
• Conos enhances the ability to study cell type recurrence and variation in complex biological systems.