Souporcell: robust clustering of single-cell RNA-seq data by genotype without reference genotypes

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) is a powerful tool for analyzing cellular heterogeneity.
• Analyzing scRNA-seq data from mixed genotypes (e.g., multiplexed donors) presents challenges in cell assignment and accurate quantification.
• Existing methods struggle to identify cross-genotype doublets and accurately estimate ambient RNA contamination.

Purpose of the Study:

• To develop a novel computational method for deconvolving scRNA-seq data from mixed genotypes.
• To accurately assign cells to their donor of origin in multiplexed experiments.
• To improve the detection of cross-genotype doublets and quantify ambient RNA contamination.

Main Methods:

• Developed 'souporcell', a method that clusters cells based on genotype variants detected in scRNA-seq reads.
• Utilized genetic variants within scRNA-seq data for cell clustering and donor assignment.
• Applied souporcell to assess genotype clustering accuracy, doublet detection, and ambient RNA estimation.

Main Results:

• Souporcell demonstrates high accuracy in clustering cells according to their genotype.
• The method effectively identifies cross-genotype doublets, even those with similar transcriptional profiles.
• Souporcell accurately estimates the level of ambient RNA contamination in scRNA-seq samples.

Conclusions:

• Souporcell provides a robust solution for deconvolving scRNA-seq data from mixed genetic origins.
• The method enhances the reliability of scRNA-seq analysis by improving doublet detection and ambient RNA correction.
• Souporcell is effective across various challenging experimental scenarios, offering a valuable tool for researchers.