The Human Cell Atlas: towards a first draft atlas

In a collection of research articles and related content, the Human Cell Atlas consortium presents tools, data and ideas towards the generation of their first draft atlas of cells in the human body

Coming less than a decade after its launch, the studies emerging from the global project are a major achievement. Funders should sign up for the long haul.

The Human Cell Atlas is yielding detailed maps of human tissues and systems throughout life, along with methods to handle single-cell data. Four scientists reflect on how the project is transforming our understanding of human biology.

As the international effort reaches a ‘critical mass’ of achievements, Nature highlights seven tools that are poised to enable the next set of discoveries.

This Perspective explores five ways in which cell atlases are revealing valuable biological insights, and how they are poised to provide considerable benefits in the coming years.

The Human Cell Atlas (HCA) aims to characterize cells from diverse individuals across the globe to better understand human biology. Here, the authors lay out principles and action items that have been adopted to affirm HCA’s commitment to equity so that the atlas is beneficial to all of humanity.

The human cell atlas (HCA) is intended as an exhaustive guidebook of human cell types and their properties. Here Kirby et al. outline how the HCA Ethics Working Group is working to build a solid foundation to address the complexities of data collection and sharing.

This Review provides a comprehensive and detailed discussion about how to build and use single-cell atlases.

A multiomic atlas of human embryonic joint and cranium development enables detailed analysis of bone and cartilage maturation.

The HOX gene cluster is responsible for anteroposterior axis patterning in an evolutionarily conserved manner. Here they examine HOX gene expression in human embryos and show that neural-crest derivatives retain the anatomical HOX code of their origin while also adopting the code of their destination.

Using single-cell and spatial transcriptomics, human embryonic limb development across space and time and the diversification and cross-species conservation of cells are demonstrated.

Understanding the development of the lung will inform treatments for congenital diseases and approaches for preterm infant care. Here they map human lung development using high-parametric imaging at the single-cell level to track abundance, proliferation, and spatial organization of key cell types during early gestation.

Quach and Farrell et al. report single-cell transcriptomic analysis of over 150,000 cell from 19 human fetal lung tissues and describe the temporal and spatial dynamics of epithelial lineage development. These epithelial lineage trajectories were further identified in human pluripotent stem cell-based models of lung cell differentiation.

A comprehensive multi-omics reference atlas of prenatal human skin shows that innate immune cells crosstalk with non-immune cells to perform pivotal roles in skin morphogenesis, including the formation of hair follicles.

The authors build an atlas of the human retinal development, using approximately 220,000 nuclei from 14 embryos and fetuses (8–23 weeks post-conception). The study reveals major cell classes, key transcription factors, and differences in the development of macular and peripheral retina.

Formation of the retina during development involves the coordinated action of retinal progenitor cells and their differentiated cell types, which is key for producing a functioning eye. Here the authors provide a detailed atlas of human retinal development, combining scRNA-seq and spatial transcriptomics, and identify key genetic factors that mediate retinal progenitor cell proliferation and differentiation.

Cell2fate improves RNA velocity analysis of single-cell and spatial transcriptomics data by module decomposition of realistic biophysical models of transcription dynamics.

UDA-seq incorporates a post-indexing step to enhance the throughput of droplet-based single-cell multimodal sequencing, enabling efficient large-scale single-cell analysis.

In a 23.4-million-cell atlas of 412 single-cell RNA-sequencing studies, SCimilarity query of macrophage and fibroblast profiles from interstitial lung disease reveals similar cell profiles across other fibrotic diseases and tissues.

Understanding single-cell multi-omics data requires powerful solutions. Here, authors present a data-efficient machine learning approach for paired data. It enables integration from unseen covariates and can link distal regulatory elements to promoters, presenting a computational version of HiC.

Popular Vote (popV) is a simple, ensemble popular vote approach for cell type annotation in single-cell omic data, flexibly incorporating various methods in an open-source Python framework. Across various challenging input datasets, popV offers consistent, accurate performance.

Genes2Genes is a dynamic programming framework that enables precise alignment for single-cell trajectories at the per-gene level.

UniCoord is a joint-VAE model designed to create a universal coordinate system for singlecell transcriptomic data, capturing major heterogeneities in a lower-dimensional latent space to enhance cell annotation and data augmentation.

Identifying cellular identities is crucial in single-cell transcriptomics. Here, authors show that large-scale deep learning-based cell annotation models, trained on hundreds of cross-tissue scRNA-seq datasets, enhance prediction quality for fine-grained highly related cell types and states.

Single-cell sequencing is vital for studying complex diseases but is costly. Here, authors introduce scSemiProfiler, a deep generative learning framework that infers single-cell profiles by combining bulk sequencing with single-cell data from selected samples, offering a cost-effective solution.

Benchmarking GRN inference methods remains a challenge. Here, authors present GRouNdGAN, a causal generative model that imposes a user-defined GRN in its architecture to simulate realistic single-cell data, bridging the gap between synthetic and biological data benchmarks of GRN inference methods.

A temporal label-aware metric helps visualize trajectories in longitudinal single-cell samples by leveraging the temporal locality of the batch effect

Effortless landmark detection is an unsupervised deep learning-based approach that addresses key challenges in landmark detection and image registration for accurate performance across diverse tissue imaging datasets.

STEM is a transfer-learning-based method that integrates spatial transcriptomics and scRNA-seq data to reconstruct localization at the single-cell level and reveal gene expression variation within cell types in tissues.

Human airway contains physiologically relevant yet rare cells, but their scarcity prevents thorough profiling and differentiation studies. Here the authors use single cell RNA sequencing to identify rare ionocytes and tuft cells, as well as a potential progenitor population with cytokine-guided differentiation into either the ionocytes or tuft cell lineage.

Here the authors characterise the cellular and molecular progression of lung alveolar damage in severe COVID-19 patients using integrated histopathology and cell atlassing, pinpointing a role for macrophage SPP1 signalling to vasculature in this process.

A quantitative morphological framework for the human thymus reveals the establishment of the lobular cytokine network, canonical thymocyte trajectories and thymic epithelial cell distributions in fetal and paediatric thymic development.

The study provides a comprehensive transcriptomic atlas of the human gastrointestinal tract across the lifespan, highlighting inflammation-induced changes in epithelial stem cells that alter mucosal architecture and promote further inflammation.

This analysis of single-cell RNA sequencing data from peripheral blood mononuclear cells for 474 individuals of diverse Asian ancestries in the Asian Immune Diversity Atlas links cell-type-specific splicing variation with autoimmune and inflammatory disease risk.

A single-cell study integrating data from lung tissues from patients with fatal COVID-19 from Malawi, the United States and Europe identifies shared and distinct immune and inflammatory mechanisms of response.

Here, Easter et al. generate a single-cell atlas of human periodontium including sulcular and junctional keratinocytes. Cell-cell communication analysis is used to predict keratinocyte-specific immune cell interactions.

A human neural organoid cell atlas integrating 36 single-cell transcriptomic datasets shows cell types and states and estimates transcriptomic similarity between primary and organoid counterparts, showing potential to assess organoid fidelity and facilitate protocol development.

This Pediatric Cell Atlas study analyzes temporal cortex single-nucleus RNA sequencing datasets from eight diverse donors from 4 to 50 years of age, describing gene expression dynamics over the course of brain maturation.

A single-cell transcriptomic study from the Human Cell Atlas integrating over 11 million brain cells from 70 studies uncovers differences across human brain regions and identifies rare progenitor and microglia subtypes.

A vascular cell atlas integrating single-cell data of 19 organs and tissues from 62 donors identifies angiotypic and organotypic characteristics of endothelial and mural cells.

The spatial single-cell multiomic atlas of the first trimester human placenta at molecular resolution provides a blueprint for future studies on early placental development and pregnancy.

The Human Endometrial Cell Atlas integrates single-cell transcriptomic datasets from women with and without endometriosis. Novel and known cell types are registered using spatial transcriptomics to provide a comprehensive map of the human endometrium in controls and endometriosis cases.

Through sequencing of 88,005 nuclei from the breast tissues of clinically healthy women of diverse genetic ancestry, a global breast single-nucleus atlas was developed that identifies distinct cell types and ancestry-level differences linked to epithelial and fibroblast cell states, which in turn could influence disease incidence, molecular subtypes and progression.

Haniffa and colleagues provide diagnostic aids, potential biomarkers for disease staging and therapeutic strategies for cutaneous T cell lymphoma.

Tumour-associated myeloid cells have been linked to patient outcome and treatment response in multiple cancer types. Here, the authors use deconvolution of single cell RNA-sequencing data to identify myeloid populations which are prognostic across cancer types.

Malmberg and colleagues generated a single-cell transcriptional reference map to investigate pan-cancer profiles of tumor-infiltrating natural killer cells.

Single-cell sequencing has enabled detailed analyses of the tumour microenvironment (TME). Here, the authors perform an integrative analysis of the TME using single-cell and spatial transcriptomics data from over a thousand tumours across thirty cancer types, identifying interferon-enriched community states predictive of immunotherapeutic responses.

The Muscle Aging Cell Atlas presents approximately 200,000 single-cell and single-nuclei transcriptomes from 17 human donors across different ages, uncovering mechanisms of aging in muscle stem cells, myofibers and microenvironment cells, and demonstrates parallels in mouse muscle aging.

Ultraconserved non-coding elements (UCNEs) can regulate developmental gene expression. Retinal multi-omics data integration revealed UCNEs to be candidate cis-regulatory elements during retinal development, which may be implicated in rare eye diseases.