The advent of single-cell omics, including single-cell RNA sequencing (scRNA-seq) has transformed our understanding of cellular heterogeneity in biological and pathogenic processes. However, a growing concern in this field is the potential for sex- and ancestry-related biases, particularly in terms of how diverse population groups are represented. Our proposed analysis aims to systematically address this bias by curating publicly available metadata from major single-cell omics datasets, including Human Cell Atlas (HCA) and Human Tumor Atlas Network (HTAN). This study will not only provide a comprehensive landscape of the current state of representation but also highlight potential biases that need correction. In doing so, we will pave the way for more representative study designs in future single-cell omics research, ultimately ensuring the resulting biomedical advances will equitably benefit people from diverse backgrounds.
Our proposed secondary analysis will focus on systematically curating and analyzing metadata from publicly available single-cell omics datasets, including those from the Human Cell Atlas (HCA) and the Human Tumor Atlas Network (HTAN). These studies frequently deposit data in GREI platforms, such as Zenodo and Mendeley Data. Our analysis will evaluate the distribution of ancestry and sex in these scOmics data to identify potential sampling bias. To achieve this, we will implement a robust methodology to curate metadata, statistically identify biases, and propose strategies for improving equity in future research. The three sub-aims include:
1. Metadata Curation and Standardization
The first step in our analysis will involve downloading and systematically curating metadata from public repositories (e.g., Zenodo, Mendeley Data). We will ensure that all metadata follows standardized formats to facilitate cross-study comparisons. Before further analyses, metadata will undergo QC, and we will harmonize metadata formats across datasets, ensuring that terms related to ancestry (e.g., “African ancestry,” “European ancestry”) and sex (“male,” “female,” “other”) are uniformly coded.
2. Ancestry and Sex Distribution Analyses
Once we have a curated set of metadata, we will conduct detailed statistical analyses to assess the distribution of ancestry and sex. Given the diversity of ancestry-related terms across datasets, we will categorize samples into broad ancestry groups based on geographic origin (e.g., African, East Asian, European, Latin American, and South Asian). These categories will be based on both self-reported data and, when available, genetic ancestry inferred from genotype data (if provided in the metadata). We will generate summary statistics to evaluate the distribution of ancestry and sex across datasets, including proportions of different ancestry groups and sex categories in each dataset and conducting cross-tabulations to evaluate whether certain combinations (e.g., male of European descent, female of African descent) are disproportionately represented or underrepresented.
3. Statistical Modeling to Identify Sampling Biases
To understand potential biases in study design, we will perform statistical analyses aimed at identifying deviations from expected ancestry and sex distributions. We will compare observed distributions with population-level census data and published estimates of US or global human diversity. Chi-square goodness-of-fit Tests will be employed to compare the observed distributions of ancestry and sex to background frequencies. Significant deviations from expected distributions will suggest biases in recruitment practices or sampling strategies. To facilitate interpretation, we will visualize the distribution of ancestry and sex across studies using bar charts, heatmaps, and population pyramids to clearly display underrepresentation trends.
The proposed secondary analysis will generate several key findings that will directly impact the field of single-cell omics research, with a particular focus on addressing biases in ancestry and sex representation across publicly available datasets. Specifically, the outcomes of this project will include:
Comprehensive Ancestry and Sex Distribution Analysis: We will provide a detailed representation of different ancestry groups and sexes across single-cell omics datasets from HCA and HTAN.
Identification of Systematic Sampling Biases: The project will uncover biases in study design, sampling strategies, and tissue types that disproportionately exclude certain populations.
Actionable Guidelines for Future Study Design: Based on the findings, we will provide a set of guidelines to researchers and funding agencies on how to design more inclusive and diverse scRNA-seq studies. This will ensure better representation of all population groups critical for the success of precision medicine initiatives.
To maximize the impact and accessibility of our findings, we will employ a multi-pronged dissemination strategy:
Open-Access Publications: We will submit and publish the outcomes of this project in preprint server (bioRxiv) and peer-reviewed journals that prioritize open access.
Public Data Repositories: All curated metadata and computational tools developed during the project will be deposited in open-access repositories such as Zenodo and Mendeley Data.
Collaborative Networks: We will engage with key stakeholders, including researchers from the HCA and HTAN consortia, diversity-focused research initiatives, and funding agencies. This will foster greater awareness of the biases present in current datasets and encourage collaborations aimed at addressing these disparities.
The proposed project will rigorously adhere to the FAIR (Findable, Accessible, Interoperable, and Reusable) principles by ensuring all curated datasets, metadata, and codes are easily discoverable and available in public repositories, such as Zenodo and Mendeley Data. Additionally, the project is aligned with the CARE principles by addressing the ethical considerations of ancestry and sex representation in genomic research. This includes fostering equitable access to the benefits of single-cell technologies and ensuring responsible use of curated data that reflects global human diversity.
To ensure replicability and reproducibility of our secondary analysis, we will implement a fully transparent workflow that includes extensive documentation of all steps, from data curation to analysis. All codes will be made publicly available via GitHub, with clear version control and user-friendly guides to facilitate independent replication. The open sharing will enable other researchers to apply our protocols to different datasets, ensuring the generalizability and reproducibility of our findings across different single-cell studies.
This project will make substantial contributions to the fields of genomics, single-cell biology, and precision medicine by addressing a critical issue—ancestry and sex biases in single-cell omics research. The curation of metadata across publicly available datasets will provide an unprecedented view of how different population groups are represented in high-throughput biological studies, particularly in the rapidly growing single-cell sequencing domain. By identifying underrepresented groups and highlighting gaps in diversity, the project will set a precedent for enhancing inclusivity in genomic research. Additionally, it will inform computational methods that adjust for population structure, offering methodological advancements that can be applied across scientific disciplines reliant on high-dimensional omics data. Furthermore, the findings from this study will push the boundaries of personalized medicine by making population diversity a core focus, thus ensuring that emerging diagnostic and therapeutic tools benefit a broader spectrum of patients.
The insights gained from this secondary analysis will have far-reaching implications for clinical applications, particularly in enhancing diagnosis, treatment, and prevention strategies. One of the critical outcomes will be the improvement of reference atlases, such as the Human Cell Atlas, to ensure they better represent global human diversity. These more comprehensive atlases will lead to more accurate and personalized diagnostic tools, as clinicians will have access to cellular and molecular profiles that encompass a range of ancestries and sexes. By reducing ancestry-related biases, the project will also facilitate the identification of disease mechanisms and therapeutic targets that are relevant to populations historically underrepresented in research. This could ultimately reduce health disparities, particularly for diseases where diagnosis and treatment have been less effective in certain groups due to insufficient data.
Our team consists of a diverse group of researchers united by a shared interest in advancing biomedicine through cutting-edge genomic research. The team came together when Kavitharini Saravanan, Aryan Saharan, and Catrina Yang each reached out to Kuan-lin Huang to volunteer in 2024 Spring.
Kuan-lin Huang leads the team as an Associate Professor at Icahn School of Medicine, bringing extensive expertise in genomics, AI, and statistical methods for biomedical applications. His experience spans leading large-scale genomic studies and developing AI-driven approaches to improve human health.
Kavitharini Saravanan is a Master’s student in bioinformatics at the University of Charlotte. She brings skills in data curation and computational biology, focusing on integrating large datasets and applying statistical techniques to understand biological phenomena.
Aryan Saharan, an undergraduate student at St. Louis University, contributes with experience in biomedical literature search and statistical analysis, supporting data preprocessing and developing analytical pipelines.
Catrina Yang, a medical student at the University of Oxford, brings a clinical perspective, ensuring the project aligns with medical relevance and facilitates the translation of research findings into health applications.
Through regular, weekly virtual meetings and collaborative tools, the team works closely to integrate their skills and ensure the project's success.
Key considerations for the success of this project include:
1. Comprehensive Data Curation: Ensuring access to diverse, high-quality single-cell omics datasets with detailed ancestry and sex metadata from HCA and HTAN.
2. Robust Analytical Methods: Employing statistical techniques to accurately assess and mitigate biases in ancestry and sex representation.
3. Open Access and Transparency: Making all curated data, methods, and results openly available to ensure reproducibility and support future research.
