scAI-SNP: a method for inferring ancestry from single-cell data

Area of Science:

• Genomics
• Population Genetics
• Bioinformatics

Background:

• Large-scale single-cell data initiatives like the Human Cell Atlas require accurate donor ancestry information.
• Self-reported race and ethnicity can be biased and unavailable for existing datasets.

Purpose of the Study:

• To introduce scAI-SNP, a novel computational tool for inferring donor ancestry directly from single-cell genomics data.
• To address the need for representative single-cell atlases reflecting human genetic diversity.

Main Methods:

• Trained scAI-SNP using 4.5 million ancestry-informative single-nucleotide polymorphisms (SNPs) from the 1000 Genomes Project dataset (3201 individuals, 26 populations).
• scAI-SNP computes the contribution of 26 population groups to a donor's ancestry from query single-cell data.

Main Results:

• scAI-SNP accurately infers ancestry from sparse single-cell data across diverse tissues and cell types (including cancer).
• The tool is robust and applicable to various single-cell profiling modalities like scRNA-seq and scATAC-seq.
• Demonstrated consistency in ancestry inference using matched whole-genome sequencing data.

Conclusions:

• Ensuring diverse ancestry representation in single-cell atlases is crucial for equitable health outcomes.
• scAI-SNP provides a robust method to determine ancestry from single-cell genomics data.
• Integrating ancestry information alongside race and ethnicity is vital for understanding and addressing human diversity in health.