The Human Cell Atlas: towards a first draft atlas

The Human Cell Atlas (HCA) consortium was founded in 2016 with the aim to build a biological atlas of every cell in the human body. Since the release of its white paper in 2017, the consortium has grown into a global network of more than 3,600 members in 102 countries, contributing data to 18 Biological Networks.

The HCA has now entered a phase of data integration towards the assembly of the first draft atlas. As part of this effort, they have put together a large collection of papers that exemplify and contribute to the key steps in this new phase, from the collection of data from embryonic, fetal and paediatric organs, to the development of computational tools for data analysis and integration and organ-specific integration. Together, these studies provide important foundations for the construction of the HCA’s first draft atlas.

In this immersive web feature, we highlight some of these papers published across the Nature Portfolio.

Nature: The Human Cell Atlas from a cell census to a unified foundation model

HCA in numbers

As an open global initiative, the HCA has attracted a growing number of members and institutes from across the world.

With the growth in the number of participating members, the number of publications resulting from HCA projects has increased over the years since the inauguration of the HCA.

The 18 Biological Networks

The HCA data portal currently hosts data from approximately 62 million cells collected from around 9,100 donors. To facilitate data integration, the consortium is constructing 18 HCA Biological Network Atlases, as shown in this diagram. Each network consolidates all available HCA data related to individual tissues or organs. To date, draft atlases from three networks — lung, nervous system and eye — have been assembled by HCA researchers collaborating globally with other consortia. The papers in this collection represent significant progress in assembling these Biological Network Atlases.

Diversity, equity and ethics

HCA data are freely accessible worldwide through the HCA data portal, the ‘gateway’ to HCA resources. The HCA is committed to creating an open, ethical, equitable and representative atlas for humanity that represents and benefits all. To support this endeavour, the HCA Equity Working Group collaborates with local scientists to organize virtual and in-person meetings, training, workshops and roadshows. Regional networks in Africa, Asia, Latin America and the Middle East, initiated and led by researchers in these areas, help to coordinate efforts, articulate priorities relevant to the populations they serve and ensure that the atlas as a whole serves all parts of the world. Anyone can become an HCA member by registering and agreeing to abide by the ethical standards and principles of the HCA.

Nature Communications: The commitment of the human cell atlas to humanity

The Ethics Working Group of the HCA was set up in 2018 to examine the ethical governance issues (for example, consent, data protection, material and data sharing) related to HCA studies. This group discusses legal issues around ethics that are relevant to the HCA, particularly in relation to its global reach, and is developing an ethics toolkit that can help members of the HCA community understand the ethical framework of the HCA.

Nature Communications: Data sharing ethics toolkit: The Human Cell Atlas

Developmental cell atlas

Developing skeleton of a human embryo at 8.5 post-conception weeks. This video has no sound. Credit: Alain Chédotal & Raphaël Blain, Institut de la Vision, Paris & MeLiS/UCBL/HCL, Lyon. From: A multi-omic atlas of human embryonic skeletal development.

Human skin organoid. Credit: Karl Koehler. From: A prenatal skin atlas reveals immune regulation of human skin morphogenesis.

Our understanding of human embryonic and fetal development has remained limited owing to the difficulty to access tissues and the limited tools available for their analysis. With the collection of single-cell data for embryonic, fetal and paediatric tissues, the HCA is providing unprecedented insights into the development of organs during gestation and childhood.

Human skin organoid. Credit: Karl Koehler. From: A prenatal skin atlas reveals immune regulation of human skin morphogenesis.

In two studies, scientists report multi-omic analyses of developing human joints and cranium and of prenatal human skin. The combination of single-cell profiling with spatial methods, over several developmental time points, allows to explore possible developmental trajectories of cell populations, define their microenvironments and infer possible mechanisms of disease.

Nature: A multi-omic atlas of human embryonic skeletal development

Nature: A prenatal skin atlas reveals immune regulation of human skin morphogenesis

Human skin cells collected in a circular shape with red and green colouring and nodules growing from the surface.

Human skin organoid. Credit: Karl Koehler. From: A prenatal skin atlas reveals immune regulation of human skin morphogenesis.

Computational tools for data analysis and integration

Graph representations of single-cell data from human neural organoids. From: An integrated transcriptomic cell atlas of human neural organoids.

Collecting data to build a human cell atlas is a challenge on its own, but additional challenges come with integrating, interrogating and utilizing such data. Multiple machine-learning-based approaches developed by studies within this collection are helping to overcome these challenges and to enhance the utility of HCA data.

Graph representations of single-cell data from human neural organoids. From: An integrated transcriptomic cell atlas of human neural organoids.

Cell type annotation

Cell type annotation is a core step in analysing single-cell RNA-sequencing data, which constitutes the majority of single-cell datasets produced by the HCA. scTab, a deep-learning model, has been developed for cross-tissue cell type annotation. Utilizing a voting scheme, PopV compares the predictions from eight integrated cell type annotation methods on an unannotated dataset. This pipeline not only identifies cell types annotated with high consensus among different methods, but also those with low consensus, suggesting that the annotations for these cells are uncertain and may require further manual inspection.

Nature Genetics: Consensus prediction of cell type labels with PopV

Nature Communications: scTab: Scaling cross-tissue single-cell annotation models

Graph representations of single-cell data from human neural organoids. From: An integrated transcriptomic cell atlas of human neural organoids.

Cell similarity search

Traditionally, to search for a cell in a new dataset, scientists first have to define it by a name or a signature and select datasets they think those cells may be present in. SCimilarity, a metric learning-based foundation model, takes a different approach by asking where other cells with a similar profile to a given cell or cell population can be found. Different from existing cell annotation tools, SCimilarity does not require predefining cell types before the search; instead, it searches for similar cells based on similarity between their expression profiles.

Nature: A cell atlas foundation model for scalable search of similar human cells

Graph representations of single-cell data from human neural organoids. From: An integrated transcriptomic cell atlas of human neural organoids.

Colourful computer generated patterns constructed from cell data resembling ink blots or flowers.

Graph representations of single-cell data from human neural organoids. From: An integrated transcriptomic cell atlas of human neural organoids.

Organ-specific integration

Using the data generated through several publications (within and outside this collection), and the computational tools developed for their analysis, several studies in this collection have started integrating data across HCA Biological Networks, including Organoid, Nervous System, Immune, Gut, Reproduction, Genetic diversity, Breast, Lung, and Heart and Vascular Biological Networks. These studies showcase the importance of collaborative data and tool generation to draw meaningful insights into the biology of these tissues, from their development to how they change during disease.

Nature: An integrated transcriptomic cell atlas of human neural organoids

Nature Genetics: A temporal cortex cell atlas highlights gene expression dynamics during human brain maturation

Nature: A spatial human thymus cell atlas mapped to a continuous tissue axis

Nature: Single-cell integration reveals metaplasia in inflammatory gut diseases

Nature Genetics: An integrated single-cell reference atlas of the human endometrium

Nature Medicine: Spatial multiomic landscape of the human placenta at molecular resolution

Nature Medicine: Single-nucleus chromatin accessibility and transcriptomic map of breast tissues of women of diverse genetic ancestry

Nature Medicine: Spatially resolved single-cell atlas unveils a distinct cellular signature of fatal lung COVID-19 in a Malawian population

Nature Medicine: An organotypic atlas of human vascular cells