A Bayesian method to infer copy number clones from single-cell RNA and ATAC sequencing

Area of Science:

• Genomics
• Computational Biology
• Cancer Research

Background:

• Single-cell RNA and ATAC sequencing provide insights into cellular phenotypes by examining gene expression and chromatin accessibility.
• Analyzing these cellular states within an evolutionary context of genetic clones is crucial for cancer research.

Purpose of the Study:

• To present CONGAS+, a Bayesian model designed to map single-cell RNA and ATAC profiles onto the latent space of copy number clones.
• To enable the clustering of cells into tumor subclones with similar ploidy for comparative analysis of expression and chromatin profiles.

Main Methods:

• CONGAS+ utilizes a Bayesian framework to integrate single-cell RNA and ATAC sequencing data.
• The model clusters cells based on copy number variations, identifying distinct tumor subclones.
• The framework is implemented on GPU for efficient analysis of large datasets, including thousands of cells.

Main Results:

• CONGAS+ successfully clusters cells into tumor subclones with similar ploidy, facilitating comparative analysis.
• The model demonstrates superior performance compared to single-molecule models and supports multi-omics assays.
• CONGAS+ effectively identifies complex subclonal architectures in prostate cancer, lymphoma, and basal cell carcinoma.

Conclusions:

• CONGAS+ provides a coherent mapping between ATAC and RNA profiles within identified tumor subclones.
• The framework facilitates the study of genotype-phenotype maps and their association with genomic instability.
• CONGAS+ is a scalable and high-performing tool for analyzing multi-omics single-cell data in cancer research.