Cell-type references generated from collections of single-cell RNA sequencing data can accelerate the functional characterization of diseases.Computational methods process and analyze sequencing data for a detailed characterization of cellular phenotypes.Single-cell profiling of different molecular layers can give further functional context to cell-type identity.The addition of spatial information can reveal immune cell function in tissue contexts. The immune system encompasses a large degree of phenotypic diversity and plasticity in its cell types, and more is to be uncovered. We argue that large, multiomic datasets of single-cell resolution, in conjunction with improved computational methods, will be essential to resolving immune cell identity. Existing datasets, combined with 'big data' methodologies, can serve as a platform to support future studies in immunology. Technical and analytical advances in multiomics and spatial integration can provide a reference for gene regulation and cellular interactions in spatially structured tissue contexts. We posit that these developments may allow guided functional studies of immune cell populations and lay the groundwork for informed cell engineering and precision medicine. The immune system encompasses a large degree of phenotypic diversity and plasticity in its cell types, and more is to be uncovered. We argue that large, multiomic datasets of single-cell resolution, in conjunction with improved computational methods, will be essential to resolving immune cell identity. Existing datasets, combined with 'big data' methodologies, can serve as a platform to support future studies in immunology. Technical and analytical advances in multiomics and spatial integration can provide a reference for gene regulation and cellular interactions in spatially structured tissue contexts. We posit that these developments may allow guided functional studies of immune cell populations and lay the groundwork for informed cell engineering and precision medicine. The human immune system is one of the most complex; further understanding these complexities can have a significant impact on preventing and curing a variety of diseases. A large number of cell types and states, many of which remain to be further characterized, underlie the many types of immune responses. Many gene products have also been studied over the years; however, owing to the low number of available high-throughput approaches, many more are either unstudied or have undetermined functions. We discuss here the most recent developments in single-cell technology and analysis, and what they can mean for immunology. Advances in single-cell RNA sequencing (scRNA-seq; Glossary) data analyses are poised to result in a complete census of human cell types [1Svensson V. et al.Exponential scaling of single-cell RNA-seq in the past decade.Nat. Protoc. 2018; 13: 599-604Crossref PubMed Scopus (388) Google Scholar, 2Svensson V. da Veiga Beltrame E. A curated database reveals trends in single cell transcriptomics.bioRxiv. 2019; (Published online August 21, 2019)https://doi.org/10.1101/742304Crossref Google Scholar]. This growth in datasets has been accompanied by the development of experimental methods that capture the ‘states’ of different molecules (DNA, RNA, and protein) in individual cells, revealing many of the regulatory underpinnings of cellular immunity, as well as virulence mechanisms of pathogens. Methods for whole-transcriptome spatial mapping are also emerging and reaching single-cell resolution, enabling for the first time the construction of an atlas of cellular interactions in complex tissues during health and disease. Supporting these advances have been developments in computational methods. In particular, the wide adoption of cutting-edge machine learning and artificial intelligence methods are set to improve our modeling and predictive power by deconvoluting gene expression networks and creating informative, integrated models of immune cells in disease. We therefore posit that single-cell approaches will, in the near future, be one of the tools most widely used for characterizing many aspects of the immune system. The earliest single-cell methods relied on protein expression to determine cell types and discern mechanisms underlying biology and disease [3Hardy R.R. et al.B-cell subpopulations identified by two-colour fluorescence analysis.Nature. 1982; 297: 589-591Crossref PubMed Scopus (102) Google Scholar]. Recently, RNA has been used as a defining molecule for single-cell phenotyping; nonetheless, important information about cellular heterogeneity can still be found at the level of DNA and proteins (Figure 1A). Methylation patterns govern gene expression [4Suzuki M.M. Bird A. DNA methylation landscapes: provocative insights from epigenomics.Nat. Rev. Genet. 2008; 9: 465-476Crossref PubMed Scopus (2212) Google Scholar], and single-cell methylation profiling has been used to distinguish rare hematopoietic stem cell subpopulations [5Hui T. et al.High-resolution single-cell DNA methylation measurements reveal epigenetically distinct hematopoietic stem cell subpopulations.Stem Cell Rep. 2018; 11: 578-592Abstract Full Text Full Text PDF PubMed Scopus (48) Google Scholar]. At the single-cell level, however, open chromatin regions are easier to profile, and are associated with regulatory and active elements in the genome, which can also be used to define cell types [6ENCODE Project ConsortiumAn integrated encyclopedia of DNA elements in the human genome.Nature. 2012; 489: 57-74Crossref PubMed Scopus (11031) Google Scholar]. These are more efficiently profiled by the assay for transposase-accessible chromatin (ATAC-seq) protocol [7Buenrostro J.D. et al.Single-cell chromatin accessibility reveals principles of regulatory variation.Nature. 2015; 523: 486-490Crossref PubMed Scopus (1055) Google Scholar, 8Cusanovich D.A. et al.Multiplex single-cell profiling of chromatin accessibility by combinatorial cellular indexing.Science. 2015; 348: 910-914Crossref PubMed Scopus (624) Google Scholar]. ATAC-seq can effectively separate immune cell populations based on transcription factor motifs detected in open chromatin peaks [9Chen X. et al.A rapid and robust method for single cell chromatin accessibility profiling.Nat. Commun. 2018; 9: 5345Crossref PubMed Scopus (94) Google Scholar]. It has also been possible to obtain information on the genome conformation of single cells by using single-cell Hi-C (scHi-C) [10Nagano T. et al.Single-cell Hi-C reveals cell-to-cell variability in chromosome structure.Nature. 2013; 502: 59-64Crossref PubMed Scopus (947) Google Scholar]. Histone modifications in individual cells have only recently been effectively profiled [11Kaya-Okur H.S. et al.CUT&Tag for efficient epigenomic profiling of small samples and single cells.Nat. Commun. 2019; 10: 30Crossref PubMed Scopus (541) Google Scholar, 12Wang Q. et al.CoBATCH for high-throughput single-cell epigenomic profiling.Mol. Cell. 2019; (Published online August 27, 2019)https://doi.org/10.1016/j.molcel.2019.07.015Abstract Full Text Full Text PDF Scopus (99) Google Scholar]. This development may significantly advance the study of transcriptional regulation in different cell types. We also envisage that other sequencing methods previously performed in cells 'in bulk' might attain single-cell resolution in the not so distant future. Protein profiling in single cells has seen important advances using mass spectrometry, such as SCoPe-MS [13Budnik B. et al.SCoPE-MS: mass spectrometry of single mammalian cells quantifies proteome heterogeneity during cell differentiation.Genome Biol. 2018; 19: 161Crossref PubMed Scopus (320) Google Scholar]; however, more reliable approaches use a panel of barcoded antibodies whose signal can be amplified by sequencing [14Stoeckius M. et al.Simultaneous epitope and transcriptome measurement in single cells.Nat. Methods. 2017; 14: 865-868Crossref PubMed Scopus (1121) Google Scholar, 15Peterson V.M. et al.Multiplexed quantification of proteins and transcripts in single cells.Nat. Biotechnol. 2017; 35: 936-939Crossref PubMed Scopus (437) Google Scholar]. An exciting example is cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq), which has greatly improved the identification of known immune subsets by combining scRNA-seq with surface protein profiling [14Stoeckius M. et al.Simultaneous epitope and transcriptome measurement in single cells.Nat. Methods. 2017; 14: 865-868Crossref PubMed Scopus (1121) Google Scholar]. To properly understand cellular mechanics, it is necessary to combine multiple measurements from RNA, DNA, and protein (Figure 1B). Integrating these molecular layers can show how regulatory networks in cells contribute to shaping the immune system. Methods have been developed for using single-cell sequencing data to infer these networks [16Aibar S. et al.SCENIC: single-cell regulatory network inference and clustering.Nat. Methods. 2017; 14: 1083-1086Crossref PubMed Scopus (1307) Google Scholar, 17Pliner H.A. et al.Cicero predicts cis-regulatory DNA interactions from single-cell chromatin accessibility data.Mol. Cell. 2018; 71: 858-871Abstract Full Text Full Text PDF PubMed Scopus (235) Google Scholar, 18Papili Gao N. et al.SINCERITIES: inferring gene regulatory networks from time-stamped single cell transcriptional expression profiles.Bioinformatics. 2018; 34: 258-266Crossref Scopus (66) Google Scholar] and to integrate the different molecular profiles of single cells [19Argelaguet R. et al.Multi-omics factor analysis – a framework for unsupervised integration of multi-omics data sets.Molecular Systems Biology. 2018; 14: e8124Crossref PubMed Scopus (333) Google Scholar, 20Angermueller C. et al.DeepCpG: accurate prediction of single-cell DNA methylation states using deep learning.Genome Biol. 2017; 18: 67Crossref PubMed Scopus (250) Google Scholar]. Combining these with pseudotime inference can inform on the regulation of dynamic processes in immunology, such as infections and development. Single-cell CRISPR/Cas9 screens can help us to learn about variation and robustness in cellular responses. These have been used to dissect T cell receptor (TCR) signaling and response to lipopolysaccharide (LPS) in dendritic cells [21Datlinger P. et al.Pooled CRISPR screening with single-cell transcriptome readout.Nat. Methods. 2017; 14: 297-301Crossref PubMed Scopus (411) Google Scholar, 22Dixit A. et al.Perturb-Seq: dissecting molecular circuits with scalable single-cell RNA profiling of pooled genetic screens.Cell. 2016; 167: 1853-1866Abstract Full Text Full Text PDF PubMed Scopus (659) Google Scholar]. The use of these technologies is still in its infancy, but we predict that they will be key in elucidating the molecular mechanisms behind complex diseases. CRISPR/Cas9 screens have been accompanied by significant breakthroughs in computational analysis, and further combination with DNA or protein profiling should propel the development and validation of causal inference methods [23Qiu X. et al.Towards inferring causal gene regulatory networks from single cell expression measurements.bioRxiv. 2018; (Published online September 25, 2018)https://doi.org/10.1101/426981Crossref Google Scholar], yielding interpretable and actionable models of immunobiology. Since the first scRNA-seq study on five mouse blastomeres [24Tang F. et al.mRNA-Seq whole-transcriptome analysis of a single cell.Nat. Methods. 2009; 6: 377-382Crossref PubMed Scopus (1877) Google Scholar], the use of single-cell sequencing technologies has seen exponential growth [1Svensson V. et al.Exponential scaling of single-cell RNA-seq in the past decade.Nat. Protoc. 2018; 13: 599-604Crossref PubMed Scopus (388) Google Scholar]. scRNA-seq is currently the method of preference to define cell states, study developmental trajectories, and characterize unknown cell populations. The rapid acquisition of large datasets surveying multiple organs [25Tabula Muris Consortium et al.Single-cell transcriptomics of 20 mouse organs creates a Tabula Muris.Nature. 2018; 562: 367-372Crossref PubMed Scopus (998) Google Scholar, 26Han X. et al.Mapping the mouse cell atlas by microwell-seq.Cell. 2018; 173: 1307Abstract Full Text Full Text PDF PubMed Scopus (129) Google Scholar], from different organisms [27Cao J. et al.Comprehensive single-cell transcriptional profiling of a multicellular organism.Science. 2017; 357: 661-667Crossref PubMed Scopus (8) Google Scholar] and at different stages of development [28Cao J. et al.The single-cell transcriptional landscape of mammalian organogenesis.Nature. 2019; 566: 496-502Crossref PubMed Scopus (967) Google Scholar], can allow us to perform informed experimental designs to answer outstanding questions in the field of immunobiology. This increase in throughput has been achieved partly thanks to a reduction in sequencing costs, but mostly due to improvements in cost-effective cell isolation (discussed in detail in [1Svensson V. et al.Exponential scaling of single-cell RNA-seq in the past decade.Nat. Protoc. 2018; 13: 599-604Crossref PubMed Scopus (388) Google Scholar]). Experimental innovations such as cell hashing [29Stoeckius M. et al.Cell Hashing with barcoded antibodies enables multiplexing and doublet detection for single cell genomics.Genome Biol. 2018; 19: 224Crossref PubMed Scopus (323) Google Scholar] and split-pool approaches [30Rosenberg A.B. et al.Single-cell profiling of the developing mouse brain and spinal cord with split-pool barcoding.Science. 2018; 360: 176-182Crossref PubMed Scopus (541) Google Scholar, 31Kang H.M. et al.Multiplexed droplet single-cell RNA-sequencing using natural genetic variation.Nat. Biotechnol. 2018; 36: 89-94Crossref PubMed Scopus (337) Google Scholar] can enable significant increases in the number of cells and donors profiled with scRNA-seq. Multidonor designs hold the promise of linking cell type-specific expression to specific diseases or variants, as recently reported in human blood cells [32van der Wijst M.G.P. et al.Single-cell RNA sequencing identifies cell type-specific cis-eQTLs and co-expression QTLs.Nat. Genet. 2018; 50: 493-497Crossref PubMed Scopus (141) Google Scholar]. As more single-cell studies move towards unraveling cell-specific responses in the immune system, cell-type annotation has been facilitated by computational methods matching cell populations across samples, tissues, and species [33Kiselev V.Y. et al.Scmap: projection of single-cell RNA-seq data across data sets.Nat. Methods. 2018; 15: 359-362Crossref PubMed Scopus (277) Google Scholar] (Figure 2A); some classifiers, such as Moana [33Kiselev V.Y. et al.Scmap: projection of single-cell RNA-seq data across data sets.Nat. Methods. 2018; 15: 359-362Crossref PubMed Scopus (277) Google Scholar] and Garnett [34Wagner F. Yanai I. Moana: a robust and scalable cell type classification framework for single-cell RNA-Seq data.bioRxiv. 2018; (Published online October 30, 2018)https://doi.org/10.1101/456129Crossref Scopus (0) Google Scholar, 35Pliner H.A. et al.Supervised classification enables rapid annotation of cell atlases.bioRxiv. 2019; (Published online February 25, 2019)https://doi.org/10.1101/538652Crossref Google Scholar], have added a layer of hierarchical stratification of cellular identity [34Wagner F. Yanai I. Moana: a robust and scalable cell type classification framework for single-cell RNA-Seq data.bioRxiv. 2018; (Published online October 30, 2018)https://doi.org/10.1101/456129Crossref Scopus (0) Google Scholar, 35Pliner H.A. et al.Supervised classification enables rapid annotation of cell atlases.bioRxiv. 2019; (Published online February 25, 2019)https://doi.org/10.1101/538652Crossref Google Scholar]. Recent work [36Lotfollahi M. et al.Generative modeling and latent space arithmetics predict single-cell perturbation response across cell types, studies and species.bioRxiv. 2018; (Published online December 14, 2018)https://doi.org/10.1101/478503Crossref Google Scholar] has taken the predictive approach a step further by combining variational autoencoders and latent space vector arithmetics to build computational models that are capable of predicting cell type-specific responses based on how other cells types respond to the same stimulus. This method has accurately predicted the transcriptional responses of different human peripheral blood mononuclear cells (PBMCs) to IFN-β stimulation in culture, based on gene expression variations of the remaining unrelated cell types; it has also predicted species-specific responses of phagocytes to LPS. Strategies based on connectionist systems, such as artificial neural networks (Box 1), might soon provide accurate predictive models that could potentially facilitate large-scale, transcriptome-wide studies of immune responses in silico (Figure 2C).Box 1Artificial Neural Networks for Single-Cell Data AnalysisThe large scale of recently generated single-cell datasets [28Cao J. et al.The single-cell transcriptional landscape of mammalian organogenesis.Nature. 2019; 566: 496-502Crossref PubMed Scopus (967) Google Scholar, 97Pijuan-Sala B. et al.A single-cell molecular map of mouse gastrulation and early organogenesis.Nature. 2019; 566: 490-495Crossref PubMed Scopus (313) Google Scholar] suggests that traditional analytical methods may not be enough to fully understand a system, and complementary methods to fully exploit these data may be needed. This has led researchers to apply methods from other fields such as physics, artificial intelligence, and machine learning to the study of single-cell multi-omic data (Figure 2). Most methods from artificial intelligence are derived from connectionist systems, and these include autoencoders and deep neural networks [98Zou J. et al.A primer on deep learning in genomics.Nat. Genet. 2019; 51: 12-18Crossref PubMed Scopus (323) Google Scholar].Machine learning uses pattern recognition and statistical inference algorithms for finding relationships or patterns in large collections of data with (supervised learning) or without (unsupervised learning) the use of explicit instructions [99Grabowski P. Rappsilber J. A primer on data analytics in functional genomics: how to move from data to insight?.Trends Biochem. Sci. 2019; 44: 21-32Abstract Full Text Full Text PDF PubMed Scopus (12) Google Scholar]. Supervised learning methods require a set of examples for use as training data such that the algorithm can later try to fit this model to new datasets. Unsupervised methods are applied without any previous training step and try to learn patterns in the data.Both autoencoders and neural networks can be used in a supervised or unsupervised way, and this decision should be based on the problem that the algorithm is intended to solve.One of the main problems of single-cell data is the experimental 'noise' that accompanies them, and one appealing way to deal with this issue is to 'clean' the data using an autoencoder. The autoencoder uses one or multiple datasets to identify features in the data that are common to all datasets, and assigns a probability of which feature does or does not represent the original data [100Doersch C. Tutorial on variational autoencoders.arRxiv. 2016; (Published online June 19, 2016)https://arxiv.org/abs/1606.05908Google Scholar]. The latent variable obtained by the autoencoder can then be used to reconstruct the data but without the noise or batch effects (Figure 2C).Artificial neural networks have become an attractive tool to study single-cell omic data. In a standard neural network each artificial neuron is arranged in a layer and connected to other artificial neurons within or between layers. The first layer captures different types of inputs that are then passed on to the underlying layers for data abstraction; the final layer collects these results to produce an output [87LeCun Y. et al.Deep learning.Nature. 2015; 521: 436-444Crossref PubMed Scopus (42113) Google Scholar]. Depending on the design, the neural network can have only a few or thousands of layers. In this way, an artificial neural network can be used to take multiple data inputs, such as expression values, protein abundance, and tissue localization, to identify a specific cell type (Figure 2D). The large scale of recently generated single-cell datasets [28Cao J. et al.The single-cell transcriptional landscape of mammalian organogenesis.Nature. 2019; 566: 496-502Crossref PubMed Scopus (967) Google Scholar, 97Pijuan-Sala B. et al.A single-cell molecular map of mouse gastrulation and early organogenesis.Nature. 2019; 566: 490-495Crossref PubMed Scopus (313) Google Scholar] suggests that traditional analytical methods may not be enough to fully understand a system, and complementary methods to fully exploit these data may be needed. This has led researchers to apply methods from other fields such as physics, artificial intelligence, and machine learning to the study of single-cell multi-omic data (Figure 2). Most methods from artificial intelligence are derived from connectionist systems, and these include autoencoders and deep neural networks [98Zou J. et al.A primer on deep learning in genomics.Nat. Genet. 2019; 51: 12-18Crossref PubMed Scopus (323) Google Scholar]. Machine learning uses pattern recognition and statistical inference algorithms for finding relationships or patterns in large collections of data with (supervised learning) or without (unsupervised learning) the use of explicit instructions [99Grabowski P. Rappsilber J. A primer on data analytics in functional genomics: how to move from data to insight?.Trends Biochem. Sci. 2019; 44: 21-32Abstract Full Text Full Text PDF PubMed Scopus (12) Google Scholar]. Supervised learning methods require a set of examples for use as training data such that the algorithm can later try to fit this model to new datasets. Unsupervised methods are applied without any previous training step and try to learn patterns in the data. Both autoencoders and neural networks can be used in a supervised or unsupervised way, and this decision should be based on the problem that the algorithm is intended to solve. One of the main problems of single-cell data is the experimental 'noise' that accompanies them, and one appealing way to deal with this issue is to 'clean' the data using an autoencoder. The autoencoder uses one or multiple datasets to identify features in the data that are common to all datasets, and assigns a probability of which feature does or does not represent the original data [100Doersch C. Tutorial on variational autoencoders.arRxiv. 2016; (Published online June 19, 2016)https://arxiv.org/abs/1606.05908Google Scholar]. The latent variable obtained by the autoencoder can then be used to reconstruct the data but without the noise or batch effects (Figure 2C). Artificial neural networks have become an attractive tool to study single-cell omic data. In a standard neural network each artificial neuron is arranged in a layer and connected to other artificial neurons within or between layers. The first layer captures different types of inputs that are then passed on to the underlying layers for data abstraction; the final layer collects these results to produce an output [87LeCun Y. et al.Deep learning.Nature. 2015; 521: 436-444Crossref PubMed Scopus (42113) Google Scholar]. Depending on the design, the neural network can have only a few or thousands of layers. In this way, an artificial neural network can be used to take multiple data inputs, such as expression values, protein abundance, and tissue localization, to identify a specific cell type (Figure 2D). Pairwise correspondence of datasets can be useful to dissect specific immune processes. However, systems-level insights will come from integrated cross-tissue datasets. The vast data collections that will make up the Human Cell Atlas [36Lotfollahi M. et al.Generative modeling and latent space arithmetics predict single-cell perturbation response across cell types, studies and species.bioRxiv. 2018; (Published online December 14, 2018)https://doi.org/10.1101/478503Crossref Google Scholar] will necessarily include an Immune Atlas of our species [37Regev A. et al.The human cell atlas.Elife. 2017; 6: e27041Crossref PubMed Scopus (973) Google Scholar]. Comparing novel data with inclusive references might also accelerate interpretation, allowing parallels to be immediately drawn across profiled tissues at steady-state or under disease conditions, and can eliminate the need to profile healthy subjects for disease studies. Establishing such references requires the development of global cell-identity models and the adoption of curated hierarchical cell-type annotations [38Bard J. et al.An ontology for cell types.Genome Biol. 2005; 6: R21Crossref PubMed Google Scholar, 39Meehan T.F. et al.Logical development of the cell ontology.BMC Bioinformatics. 2011; 12: 6Crossref PubMed Scopus (113) Google Scholar]. Nonetheless, immune cell phenotypes are also reflected in DNA modifications and protein expression, thus requiring computational methods to define cells beyond RNA molecule expression. Most cellular heterogeneity is reflected at the level of RNA expression, which can be used to characterize cell states based on markers and functional pathways. Nevertheless, multiple efforts have further probed the data for other features that can expand cellular phenotyping. High-throughput sequencing reads are at the base of expression measurements. Isoform analysis has also been an important parameter in transcriptomics but, aside from a small number of studies [40Song Y. et al.Single-cell alternative splicing analysis with Expedition reveals splicing dynamics during neuron differentiation.Mol. Cell. 2017; 67: 148-161Abstract Full Text Full Text PDF PubMed Scopus (95) Google Scholar, 41Gupta I. et al.Single-cell isoform RNA sequencing characterizes isoforms in thousands of cerebellar cells.Nat. Biotechnol. 2018; 36: 1197-1202Crossref Scopus (134) Google Scholar], remains understudied at the single-cell level. Even so, splicing variability can be highly informative in the context of an immune response. For instance, using logistic regression for differential expression analysis of scRNA-seq data has identified different isoforms of CD45 in human T cells [42Ntranos, V. et al. Identification of transcriptional signatures for cell types from single-cell RNA-seq. bioRxiv Published online February 14, 2018. https://doi.org/10.1101/258566.Google Scholar], and scRNA-seq using long-read sequencing methods has added more detailed information regarding the importance of splicing in cell identity and disease [41Gupta I. et al.Single-cell isoform RNA sequencing characterizes isoforms in thousands of cerebellar cells.Nat. Biotechnol. 2018; 36: 1197-1202Crossref Scopus (134) Google Scholar]. Differential detection of spliced and unspliced reads can also reflect transcriptional changes in the developmental trajectories of cells, with the assumption that unspliced transcripts are located in the nucleus and are more recently transcribed than those in the cytoplasm. This application of RNA kinetics to scRNA-seq data is termed RNA velocity [42Ntranos, V. et al. Identification of transcriptional signatures for cell types from single-cell RNA-seq. bioRxiv Published online February 14, 2018. https://doi.org/10.1101/258566.Google Scholar] and, among other uses, has been combined with pseudotime inference to confirm the direction of adaptation of murine T regulatory cells from a lymph node to a barrier tissue [43Miragaia R.J. et al.Single-cell transcriptomics of regulatory T cells reveals trajectories of tissue adaptation.Immunity. 2019; 50: 493-504Abstract Full Text Full Text PDF PubMed Scopus (213) Google Scholar]. Early approaches such as TraCeR have been devised to reconstruct expressed TCRs from scRNA-seq reads and determine cell clonality [44Stubbington M.J.T. et al.T cell fate and clonality inference from single-cell transcriptomes.Nat. Methods. 2016; 13: 329-332Crossref PubMed Scopus (280) Google Scholar]. This method has further been extended to B cells [45Lindeman I. et al.BraCeR: B-cell-receptor reconstruction and clonality inference from single-cell RNA-seq.Nat. Methods. 2018; 15: 563-565Crossref PubMed Scopus (41) Google Scholar], incorporating an additional lineage reconstruction step to account for somatic hypermutation events at the B cell receptor (BCR) locus. These methods were initially designed for full transcript sequencing approaches such as Smart-seq2 [46Picelli S. et al.Full-length RNA-seq from single cells using Smart-seq2.Nat. Protoc. 2014; 9: 171-181Crossref PubMed Scopus (1958) Google Scholar], but they can also be applied to droplet-based protocols, including 10X Genomics VDJ-seq. The combination of VDJ and RNA-seq at a large scale has given new insights into the relationship between activation and TCR sequences of clonotypes in the breast tumor microenvironment [47Azizi E. et al.Single-cell map of diverse immune phenotypes in the breast tumor microenvironment.Cell. 2018; 174: 1293-1308Abstract Full Text Full Text PDF PubMed Scopus (817) Google Scholar]. Moreover, increased resolution of TCR and BCR clonality has also been achieved by long-read sequencing, providing detailed descriptions of immune repertoires in various cancers [48Singh M. et al.High-throughput targeted long-read single cell sequencing reveals the clonal and transcriptional landscape of lymphocytes.Nat. Commun. 2019; 10: 3120Crossref PubMed Scopus (112) Google Scholar]. Ultimately, exploration of adaptive immunity repertoires can advance our understanding of the bias and selection of TCR and BCR chain pairs, and, together with single-cell profiling of antigen specificity, aid in inferring the association between sequence motifs and specific antigens, and presumably diseases [49Glanville J. et al.Identifying specificity groups in the