Clustering-independent estimation of cell abundances in bulk tissues using single-cell RNA-seq data

Area of Science:

• Computational biology
• Genomics
• Transcriptomics

Background:

• Single-cell RNA-sequencing (scRNA-seq) enables detailed cell state characterization.
• Bulk tissue gene expression deconvolution infers cell composition but is limited to discrete cell types.
• Current methods struggle with continuous cellular processes like differentiation and activation.

Purpose of the Study:

• To introduce ConDecon, a novel clustering-independent computational method for gene expression deconvolution.
• To improve phenotypic resolution and functional inference compared to existing regression-based methods.
• To enable the inference of cell likelihood within bulk tissue samples.

Main Methods:

• ConDecon utilizes single-cell data to infer the probability of each cell's presence in bulk tissue.
• The method is independent of predefined cell clusters.
• It can be applied to various data modalities, including bulk RNA-seq and ATAC-seq.

Main Results:

• ConDecon successfully deconvolutes bulk tissue data with enhanced phenotypic resolution.
• The method identified the role of neurodegenerative microglia and inflammatory pathways in pediatric ependymoma mesenchymal transformation.
• Spatial trajectories of microglial activation were characterized.

Conclusions:

• ConDecon advances gene expression deconvolution by accommodating continuous cellular states.
• The method offers improved functional and phenotypic insights into complex biological processes.
• ConDecon's applicability to diverse data types broadens its utility in biological research.