Analysis Of Single-cell RNA-seq Data Research Articles

ScRNASequest: an ecosystem of scRNA-seq analysis, visualization, and publishing

BackgroundSingle-cell RNA sequencing is a state-of-the-art technology to understand gene expression in complex tissues. With the growing amount of data being generated, the standardization and automation of data analysis are critical to generating hypotheses and discovering biological insights.ResultsHere, we present scRNASequest, a semi-automated single-cell RNA-seq (scRNA-seq) data analysis workflow which allows (1) preprocessing from raw UMI count data, (2) harmonization by one or multiple methods, (3) reference-dataset-based cell type label transfer and embedding projection, (4) multi-sample, multi-condition single-cell level differential gene expression analysis, and (5) seamless integration with cellxgene VIP for visualization and with CellDepot for data hosting and sharing by generating compatible h5ad files.ConclusionsWe developed scRNASequest, an end-to-end pipeline for single-cell RNA-seq data analysis, visualization, and publishing. The source code under MIT open-source license is provided at https://github.com/interactivereport/scRNASequest. We also prepared a bookdown tutorial for the installation and detailed usage of the pipeline: https://interactivereport.github.io/scRNAsequest/tutorial/docs/. Users have the option to run it on a local computer with a Linux/Unix system including MacOS, or interact with SGE/Slurm schedulers on high-performance computing (HPC) clusters.

Integrated single-cell RNA-seq analysis reveals the vital cell types and dynamic development signature of atherosclerosis.

Introduction: In the development of atherosclerosis, the remodeling of blood vessels is a key process involving plaque formation and rupture. So far, most reports mainly believe that macrophages, smooth muscle cells, and endothelial cells located at the intima and media of artery play the key role in this process. Few studies had focused on whether fibroblasts located at adventitia are involved in regulating disease process. Methods and results: In this study, we conducted in-depth analysis of single-cell RNA-seq data of the total of 18 samples from healthy and atherosclerotic arteries. This study combines several analysis methods including transcription regulator network, cell-cell communication network, pseudotime trajectory, gene set enrichment analysis, and differential expression analysis. We found that SERPINF1 is highly expressed in fibroblasts and is involved in the regulation of various signaling pathways. Conclusion: Our research reveals a potential mechanism of atherosclerosis, SERPINF1 regulates the formation and rupture of plaques through the Jak-STAT signaling pathway, which may provide new insights into the pathological study of disease. Moreover, we suggest that SRGN and IGKC as potential biomarkers for unstable arterial plaques.

Single-cell RNA-seq data analysis based on directed graph neural network.

Single-cell RNA sequencing (scRNA-seq) data scale surges with high-throughput sequencing technology development. However, although single-cell data analysis is a powerful tool, various issues have been reported, such as sequencing sparsity and complex differential patterns in gene expression. Statistical or traditional machine learning methods are inefficient, and the accuracy needs to be improved. The methods based on deep learning can not directly process non-Euclidean spatial data, such as cell diagrams. In this study, we have developed graph autoencoders and graph attention network for scRNA-seq analysis based on a directed graph neural network named scDGAE. Directed graph neural networks cannot only retain the connection properties of the directed graph but also expand the receptive field of the convolution operation. Cosine similarity, median L1 distance, and root-mean-squared error are used to measure the gene imputation performance of different methods with scDGAE. Furthermore, adjusted mutual information, normalized mutual information, completeness score, and Silhouette coefficient score are used to measure the cell clustering performance of different methods with scDGAE. Experiment results show that the scDGAE model achieves promising performance in gene imputation and cell clustering prediction on four scRNA-seq data sets with gold-standard cell labels. Furthermore, it is a robust framework that can be applied to general scRNA-Seq analyses.

Species-specific regulation of XIST by the JPX/FTX orthologs.

X chromosome inactivation (XCI) is an essential process, yet it initiates with remarkable diversity in various mammalian species. XIST, the main trigger of XCI, is controlled in the mouse by an interplay of lncRNA genes (LRGs), some of which evolved concomitantly to XIST and have orthologues across all placental mammals. Here, we addressed the functional conservation of human orthologues of two such LRGs, FTX and JPX. By combining analysis of single-cell RNA-seq data from early human embryogenesis with various functional assays in matched human and mouse pluripotent stem- or differentiated post-XCI cells, we demonstrate major functional differences for these orthologues between species, independently of primary sequence conservation. While the function of FTX is not conserved in humans, JPX stands as a major regulator of XIST expression in both species. However, we show that different entities of JPX control the production of XIST at various steps depending on the species. Altogether, our study highlights the functional versatility of LRGs across evolution, and reveals that functional conservation of orthologous LRGs may involve diversified mechanisms of action. These findings represent a striking example of how the evolvability of LRGs can provide adaptative flexibility to constrained gene regulatory networks.

The microglial immunoreceptor tyrosine-based motif-Syk signaling pathway is a promising target of immunotherapy for Alzheimer's disease.

The microglial immunoreceptor tyrosine-based motif-Syk signaling pathway is a promising target of immunotherapy for Alzheimer's disease.

Comparing the transcriptome of developing native and iPSC-derived mouse retinae by single cell RNA sequencing

We report the generation and analysis of single-cell RNA-Seq data (> 38,000 cells) from mouse native retinae and induced pluripotent stem cell (iPSC)-derived retinal organoids at four matched stages of development spanning the emergence of the major retinal cell types. We combine information from temporal sampling, visualization of 3D UMAP manifolds, pseudo-time and RNA velocity analyses, to show that iPSC-derived 3D retinal organoids broadly recapitulate the native developmental trajectories. However, we observe relaxation of spatial and temporal transcriptome control, premature emergence and dominance of photoreceptor precursor cells, and susceptibility of dynamically regulated pathways and transcription factors to culture conditions in retinal organoids. We demonstrate that genes causing human retinopathies are enriched in cell-type specifying genes and identify a subset of disease-causing genes with expression profiles that are highly conserved between human retinae and murine retinal organoids. This study provides a resource to the community that will be useful to assess and further improve protocols for ex vivo recapitulation and study of retinal development.

Comprehensive analysis of pyroptosis-related gene signatures for glioblastoma immune microenvironment and target therapy.

Glioblastoma (GBM) is a malignant brain tumour, but its subtypes (mesenchymal, classical, and proneural) show different prognoses. Pyroptosis is a programmed cell death relating to tumour progression, but its association with GBM is poorly understood. In this work, we collected 73 GBM samples (the Xiangya GBM cohort) and reported that pyroptosis involves tumour-microglia interaction and tumour response to interferon-gamma. GBM samples were grouped into different subtypes, cluster 1 and cluster 2, based on pyroptosis-related genes. Cluster 1 samples manifested a worse prognosis and had a more complicated immune landscape than cluster 2 samples. Single-cell RNA-seq data analysis supported that cluster 1 samples respond to interferon-gamma more actively. Moreover, the machine learning algorithm screened several potential compounds, including nutlin-3, for cluster 1 samples as a novel treatment. In vitro experiments supported that cluster 1 cell line, T98G, is more sensitive to nutlin-3 than cluster 2 cell line, LN229. Nutlin-3 can trigger oxidative stress by increasing DHCR24 expression. Moreover, pyroptosis-resistant genes were upregulated in LN229, which may participate against nutlin-3. Therefore, we hypothesis that GBM may be able to upregulate pyroptosis resistant related genes to against nutlin-3-triggered cell death. In summary, we conclude that pyroptosis highly associates with GBM progression, tumour immune landscape, and tumour response to nutlin-3.

TGF-β signaling and Creb5 cooperatively regulate Fgf18 to control pharyngeal muscle development.

The communication between myogenic cells and their surrounding connective tissues is indispensable for muscle morphogenesis. During late embryonic development in mice, myogenic progenitors migrate to discrete sites to form individual muscles. The detailed mechanism of this process remains unclear. Using mouse levator veli palatini (LVP) development as a model, we systematically investigated how a distinct connective tissue subpopulation, perimysial fibroblasts, communicates with myogenic cells to regulate mouse pharyngeal myogenesis. Using single-cell RNAseq data analysis, we identified that TGF-β signaling is a key regulator for the perimysial fibroblasts. Loss of TGF-β signaling in the neural crest-derived palatal mesenchyme leads to defects in perimysial fibroblasts and muscle malformation in the soft palate in Osr2Cre;Tgfbr1fl/fl mice. In particular, Creb5, a transcription factor expressed in the perimysial fibroblasts, cooperates with TGF-β signaling to activate expression of Fgf18. Moreover, Fgf18 supports pharyngeal muscle development in vivo and exogenous Fgf18 can partially rescue myogenic cell numbers in Osr2Cre;Tgfbr1fl/fl samples, illustrating that TGF-β-regulated Fgf18 signaling is required for LVP development. Collectively, our findings reveal the mechanism by which TGF-β signaling achieves its functional specificity in defining the perimysial-to-myogenic signals for pharyngeal myogenesis.

Transcriptomic analyses reveal neuronal specificity of Leigh syndrome associated genes.

Leigh syndrome is a rare, inherited, complex neurometabolic disorder with genetic and clinical heterogeneity. Features present in affected patients range from classical stepwise developmental regression to ataxia, seizures, tremor, and occasionally psychiatric manifestations. Currently, more than 100 monogenic causes of Leigh syndrome have been identified, yet the pathophysiology remains unknown. Here, we sought to determine the cellular specificity within the brain of all genes currently associated with Leigh syndrome. Further, we aimed to investigate potential genetic commonalities between Leigh syndrome and other disorders with overlapping clinical features. Enrichment of our target genes within the brain was evaluated with co-expression (CoExp) network analyses constructed using existing UK Brain Expression Consortium data. To determine the cellular specificity of the Leigh associated genes, we employed expression weighted cell type enrichment (EWCE) analysis of single-cell RNA-Seq data. Finally, CoExp network modules demonstrating enrichment of Leigh syndrome associated genes were then utilised for synaptic gene ontology analysis and heritability analysis. CoExp network analyses revealed that Leigh syndrome associated genes exhibit the highest levels of expression in brain regions most affected on MRI in affected patients. EWCE revealed significant enrichment of target genes in hippocampal and somatosensory pyramidal neurons and interneurons of the brain. Analysis of CoExp modules enriched with our target genes revealed preferential association with pre-synaptic structures. Heritability studies suggested some common enrichment between Leigh syndrome and Parkinson disease and epilepsy. Our findings suggest a primary mitochondrial dysfunction as the underlying basis of Leigh syndrome, with associated genes primarily expressed in neuronal cells.

Single-Cell RNAseq Data QC and Preprocessing.

The first step in single-cell RNAseq data analysis is the evaluation of the overall quality of the cell transcriptome and the preparation of the single-cell transcription data for clustering. In this chapter, we describe one of the possible approaches to perform single-cell data preprocessing for 3' end single-cell RNAseq transcriptomics data.

Single-Cell RNAseq Complexity Reduction.

An important step in single-cell RNAseq data analysis is the preparation of the single cell transcription data for cell sub-population partitioning. In this chapter, we describe how to perform complexity reduction for 3' end single-cell RNAseq transcriptomics data.

A Human iPSC Model Reveals Specific Upregulation of Linker Histone H1-0 in ETV6::RUNX1+ Preleukemia and B Cell Precursor Acute Lymphoblastic Leukemia

The ETV6::RUNX1 translocation is the most common chromosomal rearrangement associated with pediatric B cell precursor acute lymphoblastic leukemia (BCP-ALL). As previously shown, ETV6::RUNX1 leads to a distinct clinically silent preleukemic state in utero that manifests in impaired early B cell differentiation. Acquisition of secondary oncogenic mutations gives rise to overt leukemia in 0.2% of children carrying ETV6::RUNX1. In-depth characterization of ETV6::RUNX1+ preleukemia is essential to gain an understanding of leukemic transformation and develop preventive strategies. Here, we identify dysregulation of the epigenetic regulator H1-0 in ETV6::RUNX1+ preleukemia and BCP-ALL. Utilizing CRISPR/Cas9 genome editing, we generated monoclonal ETV6::RUNX1+ human induced pluripotent stem cell (hiPSC) lines with stable ETV6::RUNX1 expression driven by the endogenous ETV6 promoter. To assess the preleukemic setting mediated by ETV6::RUNX1, we performed RNA sequencing (RNA-seq) analysis of ETV6::RUNX1+ and matching wild-type (WT) hiPSCs of two donors. We identified 20 differentially expressed genes in ETV6::RUNX1+ hiPSCs of both donors compared to the respective WT hiPSCs (up: 15, down: 5, absolute fold change ≥ 2, p ≤ 0.05). Interestingly, our RNA-seq analysis revealed consistent upregulation of H1-0 (mean: 2.43-fold), a linker histone involved in chromatin remodeling, cell proliferation and differentiation. Elevated H1-0 levels in the ETV6::RUNX1+ hiPSC lines were confirmed by quantitative real-time PCR and immunoblot (Figure 1A). To determine H1-0 levels in ETV6::RUNX1+ B cell precursors, we analyzed published gene expression microarray and RNA-seq data sets. Both ETV6::RUNX1+ CD34+ cells (n = 10) derived from umbilical cord blood (Polak et al. Haematologica 2019) and ETV6::RUNX1+ CD34+CD19+ cells (n = 4) generated by hematopoietic differentiation of hiPSCs (Böiers et al. Developmental Cell 2018) showed significant upregulation of H1-0 compared to the respective control cells lacking ETV6::RUNX1 (unpaired t-test, p = 0.0087 and p = 0.0076, respectively). Additionally, we differentiated ETV6::RUNX1+ and WT hiPSCs into CD34+CD45+ hematopoietic progenitor cells and performed single-cell RNA-seq analysis. We identified coexpression of H1-0 with cyclin-dependent kinase inhibitor genes CDKN1A and CDKN1B, indicating a slow or non-cycling phenotype of H1-0+ cells. Next, we examined H1-0 mRNA levels in gene expression data sets of pediatric and adult leukemia patients (St. Jude PeCan database (n = 876), Jerchel et al. Haematologica 2018 (n = 654), MILE study (n = 1299)). H1-0 expression was found to be significantly elevated in ETV6::RUNX1+ BCP-ALL patient samples compared to other leukemia entities, especially TCF3::PBX1+ ALL (ordinary one-way ANOVA, p ≤ 0.0001), KMT2A rearranged ALL (p ≤ 0.0001) and chronic lymphoblastic leukemia (CLL, p ≤ 0.0001). Strikingly, H1-0 expression was 22.37-fold lower in CLL compared to ETV6::RUNX1+ BCP-ALL (Figure 1B). This observation prompted us to investigate H1-0 expression in B cell developmental stages using published data sets. Interestingly, we found that H1-0 levels decrease during early B cell development in both normal bone marrow (Black et al. Nucleic Acids Research 2018) and umbilical cord blood-derived cells (Novershtern et al. Cell 2011). Additionally, analysis of single-cell RNA-seq data of early B cell precursors derived from healthy bone marrow (Mehtonen et al. Genome Medicine 2020) and fetal liver (Developmental Cell Atlas Newcastle University) revealed that relative numbers of H1-0 expressing cells per differentiation state decreased along the pseudotime B cell development trajectory. Taken together, our study identifies H1-0 upregulation as a specific alteration present already at the ETV6::RUNX1+ preleukemic state. High H1-0 expression was associated with slower proliferation and increased stemness. This phenotype might explain the persistence of in utero generated preleukemic ETV6::RUNX1+ clones residing in the bone marrow of approximately 5% of healthy children for years. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

ScGNN 2.0: a graph neural network tool for imputation and clustering of single-cell RNA-Seq data.

Gene expression imputation has been an essential step of the single-cell RNA-Seq data analysis workflow. Among several deep-learning methods, the debut of scGNN gained substantial recognition in 2021 for its superior performance and the ability to produce a cell-cell graph. However, the implementation of scGNN was relatively time-consuming and its performance could still be optimized. The implementation of scGNN 2.0 is significantly faster than scGNN thanks to a simplified close-loop architecture. For all eight datasets, cell clustering performance was increased by 85.02% on average in terms of adjusted rand index, and the imputation Median L1 Error was reduced by 67.94% on average. With the built-in visualizations, users can quickly assess the imputation and cell clustering results, compare against benchmarks and interpret the cell-cell interaction. The expanded input and output formats also pave the way for custom workflows that integrate scGNN 2.0 with other scRNA-Seq toolkits on both Python and R platforms. scGNN 2.0 is implemented in Python (as of version 3.8) with the source code available at https://github.com/OSU-BMBL/scGNN2.0. Supplementary data are available at Bioinformatics online.

Variant calling enhances the identification of cancer cells in single-cell RNA sequencing data.

Single-cell RNA-sequencing is an invaluable research tool that allows for the investigation of gene expression in heterogeneous cancer cell populations in ways that bulk RNA-seq cannot. However, normal (i.e., non tumor) cells in cancer samples have the potential to confound the downstream analysis of single-cell RNA-seq data. Existing methods for identifying cancer and normal cells include copy number variation inference, marker-gene expression analysis, and expression-based clustering. This work aims to extend the existing approaches for identifying cancer cells in single-cell RNA-seq samples by incorporating variant calling and the identification of putative driver alterations. We found that putative driver alterations can be detected in single-cell RNA-seq data obtained with full-length transcript technologies and noticed that a subset of cells in tumor samples are enriched for putative driver alterations as compared to normal cells. Furthermore, we show that the number of putative driver alterations and inferred copy number variation are not correlated in all samples. Taken together, our findings suggest that augmenting existing cancer-cell filtering methods with variant calling and analysis can increase the number of tumor cells that can be confidently included in downstream analyses of single-cell full-length transcript RNA-seq datasets.

Self-supervised contrastive learning for integrative single cell RNA-seq data analysis.

We present a novel self-supervised Contrastive LEArning framework for single-cell ribonucleic acid (RNA)-sequencing (CLEAR) data representation and the downstream analysis. Compared with current methods, CLEAR overcomes the heterogeneity of the experimental data with a specifically designed representation learning task and thus can handle batch effects and dropout events simultaneously. It achieves superior performance on a broad range of fundamental tasks, including clustering, visualization, dropout correction, batch effect removal, and pseudo-time inference. The proposed method successfully identifies and illustrates inflammatory-related mechanisms in a COVID-19 disease study with 43 695 single cells from peripheral blood mononuclear cells.

SCDD: a novel single-cell RNA-seq imputation method with diffusion and denoising.

Single-cell sequencing technologies are widely used to discover the evolutionary relationships and the differences in cells. Since dropout events may frustrate the analysis, many imputation approaches for single-cell RNA-seq data have appeared in previous attempts. However, previous imputation attempts usually suffer from the over-smooth problem, which may bring limited improvement or negative effect for the downstream analysis of single-cell RNA-seq data. To solve this difficulty, we propose a novel two-stage diffusion-denoising method called SCDD for large-scale single-cell RNA-seq imputation in this paper. We introduce the diffusion i.e. a direct imputation strategy using the expression of similar cells for potential dropout sites, to perform the initial imputation at first. After the diffusion, a joint model integrated with graph convolutional neural network and contractive autoencoder is developed to generate superposition states of similar cells, from which we restore the original states and remove the noise introduced by the diffusion. The final experimental results indicate that SCDD could effectively suppress the over-smooth problem and remarkably improve the effect of single-cell RNA-seq downstream analysis, including clustering and trajectory analysis.

Comprehensive Analysis of the Expression and Prognosis for Tripartite Motif-Containing Genes in Breast Cancer.

Tripartite motif-containing genes (TRIMs), with a ubiquitin ligase’s function, play critical roles in antitumor immunity by activating tumor-specific immune responses and stimulating tumor proliferation, thus affecting patient outcomes. However, the expression pattern and prognostic values of TRIMs in breast cancer (BC) are not well clarified. In this study, several datasets and software were integrated to perform a comprehensive analysis of the expression pattern in TRIMs and investigate their prognosis values in BC. We found that TRIM59/46 were significantly upregulated and TRIM66/52-AS1/68/7/2/9/29 were decreased in BC and validated them using an independent cohort. The expression of numerous TRIMs are significantly correlated with BC molecular subtypes, but not with tumor stages or patient age at diagnosis. Higher expression of TRIM3/14/69/45 and lower expressions of TRIM68/2 were associated with better overall survival in BC using the Kaplan–Meier analysis. The multivariate Cox proportional hazards model identified TRIM45 as an independent prognostic marker. Further analysis of single-cell RNA-seq data revealed that most TRIMs are also expressed in nontumor cells. Higher expression of some TRIMs in the immune or stromal cells suggests an important role of TRIMs in the BC microenvironment. Functional enrichment of the co-expression genes indicates that they may be involved in muscle contraction and interferon-gamma signaling pathways. In brief, through the analysis, we provided several TRIMs that may contribute to the tumor progression and TRIM45 as a potential new prognostic biomarker for BC.

Differential Expression Analysis of Single-Cell RNA-Seq Data: Current Statistical Approaches and Outstanding Challenges.

With the advent of single-cell RNA-sequencing (scRNA-seq), it is possible to measure the expression dynamics of genes at the single-cell level. Through scRNA-seq, a huge amount of expression data for several thousand(s) of genes over million(s) of cells are generated in a single experiment. Differential expression analysis is the primary downstream analysis of such data to identify gene markers for cell type detection and also provide inputs to other secondary analyses. Many statistical approaches for differential expression analysis have been reported in the literature. Therefore, we critically discuss the underlying statistical principles of the approaches and distinctly divide them into six major classes, i.e., generalized linear, generalized additive, Hurdle, mixture models, two-class parametric, and non-parametric approaches. We also succinctly discuss the limitations that are specific to each class of approaches, and how they are addressed by other subsequent classes of approach. A number of challenges are identified in this study that must be addressed to develop the next class of innovative approaches. Furthermore, we also emphasize the methodological challenges involved in differential expression analysis of scRNA-seq data that researchers must address to draw maximum benefit from this recent single-cell technology. This study will serve as a guide to genome researchers and experimental biologists to objectively select options for their analysis.

Embigin is a fibronectin receptor that affects sebaceous gland differentiation and metabolism.

Stem cell renewal and differentiation are regulated by interactions with the niche. Although multiple cell populations have been identified in distinct anatomical compartments, little is known about niche-specific molecular factors. Using skin as a model system and combining single-cell RNA-seq data analysis, immunofluorescence, and transgenic mouse models, we show that the transmembrane protein embigin is specifically expressed in the sebaceous gland and that the number of embigin-expressing cells is negatively regulated by Wnt. The loss of embigin promotes exit from the progenitor compartment and progression toward differentiation, and also compromises lipid metabolism. Embigin modulates sebaceous niche architecture by affecting extracellular matrix organization and basolateral targeting of monocarboxylate transport. We discover through ligand screening that embigin is a direct fibronectin receptor, binding to the N-terminal fibronectin domain without impairing integrin function. Our results solve the long-standing question of how embigin regulates cell adhesion and demonstrate a mechanism that couples adhesion and metabolism.

DGCyTOF: Deep learning with graphic cluster visualization to predict cell types of single cell mass cytometry data.

Single-cell mass cytometry, also known as cytometry by time of flight (CyTOF) is a powerful high-throughput technology that allows analysis of up to 50 protein markers per cell for the quantification and classification of single cells. Traditional manual gating utilized to identify new cell populations has been inadequate, inefficient, unreliable, and difficult to use, and no algorithms to identify both calibration and new cell populations has been well established. A deep learning with graphic cluster (DGCyTOF) visualization is developed as a new integrated embedding visualization approach in identifying canonical and new cell types. The DGCyTOF combines deep-learning classification and hierarchical stable-clustering methods to sequentially build a tri-layer construct for known cell types and the identification of new cell types. First, deep classification learning is constructed to distinguish calibration cell populations from all cells by softmax classification assignment under a probability threshold, and graph embedding clustering is then used to identify new cell populations sequentially. In the middle of two-layer, cell labels are automatically adjusted between new and unknown cell populations via a feedback loop using an iteration calibration system to reduce the rate of error in the identification of cell types, and a 3-dimensional (3D) visualization platform is finally developed to display the cell clusters with all cell-population types annotated. Utilizing two benchmark CyTOF databases comprising up to 43 million cells, we compared accuracy and speed in the identification of cell types among DGCyTOF, DeepCyTOF, and other technologies including dimension reduction with clustering, including Principal Component Analysis (PCA), Factor Analysis (FA), Independent Component Analysis (ICA), Isometric Feature Mapping (Isomap), t-distributed Stochastic Neighbor Embedding (t-SNE), and Uniform Manifold Approximation and Projection (UMAP) with k-means clustering and Gaussian mixture clustering. We observed the DGCyTOF represents a robust complete learning system with high accuracy, speed and visualization by eight measurement criteria. The DGCyTOF displayed F-scores of 0.9921 for CyTOF1 and 0.9992 for CyTOF2 datasets, whereas those scores were only 0.507 and 0.529 for the t-SNE+k-means; 0.565 and 0.59, for UMAP+ k-means. Comparison of DGCyTOF with t-SNE and UMAP visualization in accuracy demonstrated its approximately 35% superiority in predicting cell types. In addition, observation of cell-population distribution was more intuitive in the 3D visualization in DGCyTOF than t-SNE and UMAP visualization. The DGCyTOF model can automatically assign known labels to single cells with high accuracy using deep-learning classification assembling with traditional graph-clustering and dimension-reduction strategies. Guided by a calibration system, the model seeks optimal accuracy balance among calibration cell populations and unknown cell types, yielding a complete and robust learning system that is highly accurate in the identification of cell populations compared to results using other methods in the analysis of single-cell CyTOF data. Application of the DGCyTOF method to identify cell populations could be extended to the analysis of single-cell RNASeq data and other omics data.