Related Experiment Video
Updated: Aug 6, 2026
09:06
Spatially Resolved, Integrated Single-Cell Multiomic Profiling of the Transcriptome and Epigenomic Targets in Frozen Tissue Sections
Published on: June 12, 2026
Are different populations fairly represented in single-cell omic atlases?
Catrina Yang1, Kavitharini Saravanan2, Aryan Saharan3
1University of Oxford, Green Templeton College, Medical Sciences Division, Oxford OX2 6HG, UK.
Cell Genomics
|July 20, 2026
Summary
Single-cell atlases show significant demographic bias, overrepresenting European individuals and underrepresenting Asian and Latino populations. This inequity risks embedding bias into AI models and therapeutic development.
Area of Science:
- Genomics
- Biomedical Research
- Population Health
Background:
- Single-cell omic atlases are crucial for advancing biology and medicine.
- Systematic evaluation of demographic representativeness in these atlases is lacking.
- Existing atlases may not reflect global population diversity.
Purpose of the Study:
- To systematically evaluate the demographic representativeness of major single-cell omic atlases.
- To identify and quantify population-level disparities in sample representation.
- To highlight the implications of these disparities for AI, biomarker discovery, and therapeutics.
Main Methods:
- Analysis of >13,500 samples from Human Cell Atlas (HCA), Human Tumor Atlas Network (HTAN), and PsychAD Consortium.
- Benchmarking atlas demographics against global and US general and disease-prevalence data.
- Assessment of ancestry and sex representation within annotated samples.
Main Results:
- Pervasive overrepresentation of European ancestry and underrepresentation of Asian and Latino individuals across analyzed atlases.
- Significant lack of ancestry annotation in HCA samples (nearly 70%).
- European overrepresentation observed in HCA (6-fold), PsychAD (two-thirds), and HTAN (69%), with sex skews in HTAN tumors.
Conclusions:
- Current single-cell atlases exhibit significant demographic inequities, primarily favoring European ancestry.
- These disparities pose a risk of embedding biases into AI foundational models, biomarker discovery, and therapeutic development.
- Actionable strategies and a checklist are proposed to promote equitable design in future single-cell studies.
