Background Large-scale single-cell transcriptomic datasets generated using different technologies contain batch-specific systematic variations that present a challenge to batch-effect removal and data integration. With continued growth expected in scRNA-seq data, achieving effective batch integration with available computational resources is crucial. Here, we perform an in-depth benchmark study on available batch correction methods to determine the most suitable method for batch-effect removal. Results We compare 14 methods in terms of computational runtime, the ability to handle large datasets, and batch-effect correction efficacy while preserving cell type purity. Five scenarios are designed for the study: identical cell types with different technologies, non-identical cell types, multiple batches, big data, and simulated data. Performance is evaluated using four benchmarking metrics including kBET, LISI, ASW, and ARI. We also investigate the use of batch-corrected data to study differential gene expression. Conclusion Based on our results, Harmony, LIGER, and Seurat 3 are the recommended methods for batch integration. Due to its significantly shorter runtime, Harmony is recommended as the first method to try, with the other methods as viable alternatives.

Macrophages promote both injury and repair following myocardial infarction, but discriminating functions within mixed populations remains challenging. Here we used fate mapping and single-cell transcriptomics to demonstrate that at steady state, TIMD4 + LYVE1 + MHC-II lo CCR2 − resident cardiac macrophages self-renew with negligible blood monocyte input. Monocytes partially replaced resident TIMD4 − LYVE1 − MHC-II hi CCR2 − macrophages and fully replaced TIMD4 − LYVE1 − MHC-II hi CCR2 + macrophages, revealing a hierarchy of monocyte contribution to functionally distinct macrophage subsets. Ischemic injury reduced TIMD4 + and TIMD4 − resident macrophage abundance within infarcted tissue while recruited, CCR2 + monocyte-derived macrophages adopted multiple cell fates, including those nearly indistinguishable from resident macrophages. Despite this similarity, inducible depletion of resident macrophages using a Cx3cr1 -based system led to impaired cardiac function and promoted adverse remodeling primarily within the peri-infarct zone, highlighting a non-redundant, cardioprotective role of resident cardiac macrophages. Lastly, we demonstrate the ability of TIMD4 to be used as a durable lineage marker of a subset of resident cardiac macrophages.

Mouse conventional dendritic cells (cDCs) can be classified into two functionally distinct lineages: the CD8α(+) (CD103(+)) cDC1 lineage, and the CD11b(+) cDC2 lineage. cDCs arise from a cascade of bone marrow (BM) DC-committed progenitor cells that include the common DC progenitors (CDPs) and pre-DCs, which exit the BM and seed peripheral tissues before differentiating locally into mature cDCs. Where and when commitment to the cDC1 or cDC2 lineage occurs remains poorly understood. Here we found that transcriptional signatures of the cDC1 and cDC2 lineages became evident at the single-cell level from the CDP stage. We also identified Siglec-H and Ly6C as lineage markers that distinguished pre-DC subpopulations committed to the cDC1 lineage (Siglec-H(-)Ly6C(-) pre-DCs) or cDC2 lineage (Siglec-H(-)Ly6C(+) pre-DCs). Our results indicate that commitment to the cDC1 or cDC2 lineage occurs in the BM and not in the periphery.

Spatial transcriptomics enable us to dissect tissue heterogeneity and map out inter-cellular communications. Optimal integration of transcriptomics data and associated spatial information is essential towards fully exploiting the data. We present SEDR, an unsupervised spatially embedded deep representation of both transcript and spatial information. The SEDR pipeline uses a deep autoencoder to construct a low-dimensional latent representation of gene expression, which is then simultaneously embedded with the corresponding spatial information through a variational graph autoencoder. We applied SEDR on human dorsolateral prefrontal cortex data and achieved better clustering accuracy, and correctly retraced the prenatal cortex development order with trajectory analysis. We also found the SEDR representation to be eminently suited for batch integration. Applying SEDR to human breast cancer data, we discerned heterogeneous sub-regions within a visually homogenous tumor region, identifying a tumor core with pro-inflammatory microenvironment and an outer ring region enriched with tumor associated macrophages which drives an immune suppressive microenvironment.

Single-cell RNA-sequencing offers unprecedented resolution of the continuum of state transition during cell differentiation and development. However, tools for constructing multi-branching cell lineages from single-cell data are limited. Here we present Mpath, an algorithm that derives multi-branching developmental trajectories using neighborhood-based cell state transitions. Applied to mouse conventional dendritic cell (cDC) progenitors, Mpath constructs multi-branching trajectories spanning from macrophage/DC progenitors through common DC progenitor to pre-dendritic cells (preDC). The Mpath-generated trajectories detect a branching event at the preDC stage revealing preDC subsets that are exclusively committed to cDC1 or cDC2 lineages. Reordering cells along cDC development reveals sequential waves of gene regulation and temporal coupling between cell cycle and cDC differentiation. Applied to human myoblasts, Mpath recapitulates the time course of myoblast differentiation and isolates a branch of non-muscle cells involved in the differentiation. Our study shows that Mpath is a useful tool for constructing cell lineages from single-cell data.

Spatial transcriptomics technologies generate gene expression profiles with spatial context, requiring spatially informed analysis tools for three key tasks, spatial clustering, multisample integration, and cell-type deconvolution. We present GraphST, a graph self-supervised contrastive learning method that fully exploits spatial transcriptomics data to outperform existing methods. It combines graph neural networks with self-supervised contrastive learning to learn informative and discriminative spot representations by minimizing the embedding distance between spatially adjacent spots and vice versa. We demonstrated GraphST on multiple tissue types and technology platforms. GraphST achieved 10% higher clustering accuracy and better delineated fine-grained tissue structures in brain and embryo tissues. GraphST is also the only method that can jointly analyze multiple tissue slices in vertical or horizontal integration while correcting batch effects. Lastly, GraphST demonstrated superior cell-type deconvolution to capture spatial niches like lymph node germinal centers and exhausted tumor infiltrating T cells in breast tumor tissue.

The ability to study cellular heterogeneity at single cell resolution is making single-cell sequencing increasingly popular. However, there is no publicly available resource that offers an integrated cell atlas with harmonized metadata that users can integrate new data with. Here, we present DISCO (https://www.immunesinglecell.org/), a database of Deeply Integrated Single-Cell Omics data. The current release of DISCO integrates more than 18 million cells from 4593 samples, covering 107 tissues/cell lines/organoids, 158 diseases, and 20 platforms. We standardized the associated metadata with a controlled vocabulary and ontology system. To allow large scale integration of single-cell data, we developed FastIntegration, a fast and high-capacity version of Seurat Integration. We also developed CELLiD, an atlas guided automatic cell type identification tool. Employing these two tools on the assembled data, we constructed one global atlas and 27 sub-atlases for different tissues, diseases, and cell types. DISCO provides three online tools, namely Online FastIntegration, Online CELLiD, and CellMapper, for users to integrate, annotate, and project uploaded single-cell RNA-seq data onto a selected atlas. Collectively, DISCO is a versatile platform for users to explore published single-cell data and efficiently perform integrated analysis with their own data.

Nuclear factor of activated T cells (NFAT) comprises a family of transcription factors that regulate T cell development, activation and differentiation. NFAT signalling can also mediate granulocyte and dendritic cell (DC) activation, but it is unknown whether NFAT influences their development from progenitors. Here, we report a novel role for calcineurin/NFAT signalling as a negative regulator of myeloid haematopoiesis. Reconstituting lethally irradiated mice with haematopoietic stem cells expressing an NFAT-inhibitory peptide resulted in enhanced development of the myeloid compartment. Culturing bone marrow cells in media supplemented with Flt3-L in the presence of the calcineurin/NFAT inhibitor Cyclosporin A increased numbers of differentiated DC. Global gene expression analysis of untreated DC and NFAT-inhibited DC revealed differential expression of transcripts that regulate cell cycle and apoptosis. In conclusion, these results provide evidence that calcineurin/NFAT signalling negatively regulates myeloid lineage development. The finding that inhibition of NFAT enhances myeloid development provides a novel insight into understanding how the treatment with drugs targeting calcineurin/NFAT signalling influence the homeostasis of the innate immune system.