Single Cell Transcriptomics Implicate Novel Monocyte and T Cell Immune Dysregulation in Sarcoidosis

Sarcoidosis is a systemic inflammatory disease characterized by infiltration of immune cells into granulomas. Previous gene expression studies using heterogeneous cell mixtures lack insight into cell-type-specific immune dysregulation. We performed the first single-cell RNA-sequencing study of sarcoidosis in peripheral immune cells in 48 patients and controls. Following unbiased clustering, differentially expressed genes were identified for 18 cell types and bioinformatically assessed for function and pathway enrichment. Our results reveal persistent activation of circulating classical monocytes with subsequent upregulation of trafficking molecules. Specifically, classical monocytes upregulated distinct markers of activation including adhesion molecules, pattern recognition receptors, and chemokine receptors, as well as enrichment of immunoregulatory pathways HMGB1, mTOR, and ephrin receptor signaling. Predictive modeling implicated TGFβ and mTOR signaling as drivers of persistent monocyte activation. Additionally, sarcoidosis T cell subsets displayed patterns of dysregulation. CD4 naïve T cells were enriched for markers of apoptosis and Th17/Treg differentiation, while effector T cells showed enrichment of anergy-related pathways. Differentially expressed genes in regulatory T cells suggested dysfunctional p53, cell death, and TNFR2 signaling. Using more sensitive technology and more precise units of measure, we identify cell-type specific, novel inflammatory and regulatory pathways. Based on our findings, we suggest a novel model involving four convergent arms of dysregulation: persistent hyperactivation of innate and adaptive immunity via classical monocytes and CD4 naïve T cells, regulatory T cell dysfunction, and effector T cell anergy. We further our understanding of the immunopathology of sarcoidosis and point to novel therapeutic targets.

Download Full-text

Gene Expression Profiling of Sorted Peripheral Blood Cells Using Microarray and Next Generation Sequencing Reveals Distinct Molecular Signatures in the Polymorphonuclear and Mononuclear Cells of Patients with Polycythemia Vera and Primary Myelofibrosis

Blood ◽

10.1182/blood.v126.23.5201.5201 ◽

2015 ◽

Vol 126 (23) ◽

pp. 5201-5201

Author(s):

Chieh Lee Wong ◽

Baoshan Ma ◽

Gareth Gerrard ◽

Martyna Adamowicz-Brice ◽

Zainul Abidin Norziha ◽

...

Keyword(s):

Gene Expression ◽

T Cells ◽

Gene Expression Profiling ◽

Differential Expression ◽

Differentially Expressed Genes ◽

Expression Profiling ◽

Differentially Expressed ◽

Cell Type ◽

Cell Type Specific ◽

Generation Sequencing

Abstract Background The past decade has witnessed a significant progress in the understanding of the molecular pathogenesis of myeloproliferative neoplasms (MPN). A large number of genes have now been implicated in the pathogenesis of MPN but their relative importance, the mechanisms by which they cause different cell types to predominate and their implications for prognosis remain unknown. We hypothesized that there are other genes which may contribute to the pathogenesis of the different disease subtypes detectable only by cell-type specific analysis. Aim The aim of this study was to perform gene expression profiling on different cell types from patients with MPN in order to identify novel variants and driver mutations, to elucidate the pathogenesis and to identify predictors of survival in patients with MPN in a multiracial country. Methods We performed gene expression profiling on normal controls (NC) and patients with MPN from 3 different races (Malay, Chinese and Indian) in Malaysia who were diagnosed with essential thrombocythemia (ET), polycythemia vera (PV) and primary myelofibrosis (PMF) according to the 2008 WHO diagnostic criteria for MPN. Two cohorts of patients, the patient and validation cohorts, from 3 tertiary-level hospitals were recruited prospectively over 3 years and informed consents were obtained. Peripheral blood samples were taken and sorted into polymorphonuclear cells (PMNs), mononuclear cells (MNCs) and T cells. RNA was extracted from each cell population. Gene expression profiling was performed using the Illumina HumanHT-12 Expression Beadchip for microarray and the Illumina Nextera XT DNA Sample Preparation Kit for next generation sequencing on the patient and validation cohorts respectively. Results Twenty-eight patients (10 ET, 11 PV and 7 PMF) and 11 NC were recruited into the patient cohort. Twelve patients (4 ET, 4 PV and 4 PMF) and 4 NC were recruited into the validation cohort. Gene expression levels for each cell type in each disease were compared with NC. In the patient cohort, the number of differentially expressed genes in ET, PV and PMF was 0, 141 and 15 respectively for PMNs (p < 0.05 after multiple testing correction) and 5, 170 and 562 respectively for MNCs (p < 0.05). No differentially expressed genes were identified for T cells in any of the three disease groups. RNA-seq analysis of samples from the validation cohort was used to corroborate these findings. After combination, we were able to confirm differential expression of 0, 14 and 7 genes in ET, PV and PMF respectively for PMNs (p < 0.05) and 51 genes in only PMF for MNCs (p < 0.05). The validated differentially expressed genes for PMNs and MNCs were mutually exclusive except for one gene. The differentially expressed genes in PV and PMF for PMNs were involved in cellular processes and metabolic pathways whereas the differentially expressed genes for PMF in MNCs were involved in regulation of cytoskeleton, focal adhesion and cell signaling pathways. Conclusion This is the first study to use microarray and next generation sequencing techniques to compare cell type-specific expression of genes between different subtypes of MPN. The lack of differential expression in T cells validates the techniques used and indicates that they are not part of the neoplastic clone. Differential expression of genes for MNCs was seen only in PMF which may be related to their more severe phenotype. Interestingly, there were fewer differentially expressed genes in PMF compared to PV for PMNs. The lack of differential expression in ET may either reflect the relatively milder phenotype of the disease or that differential expression is limited to megakaryocytes-platelets which were not studied. The lists of mutually exclusive cell type-specific differentially expressed genes for PMNs and MNCs provide further insight into the pathogenesis of MPN and into the differences between its different forms. The identified genes also indicate further routes for investigation of pathogenesis and possible disease-specific targets for therapy. Disclosures Aitman: Illumina: Honoraria.

Download Full-text

Confronting false discoveries in single-cell differential expression

Nature Communications ◽

10.1038/s41467-021-25960-2 ◽

2021 ◽

Vol 12 (1) ◽

Author(s):

Jordan W. Squair ◽

Matthieu Gautier ◽

Claudia Kathe ◽

Mark A. Anderson ◽

Nicholas D. James ◽

...

Keyword(s):

Single Cell ◽

Differential Expression ◽

Differentially Expressed Genes ◽

Differential Expression Analysis ◽

Differentially Expressed ◽

Cell Type ◽

Mouse Spinal Cord ◽

Cell Type Specific ◽

False Discoveries ◽

Biological Differences

AbstractDifferential expression analysis in single-cell transcriptomics enables the dissection of cell-type-specific responses to perturbations such as disease, trauma, or experimental manipulations. While many statistical methods are available to identify differentially expressed genes, the principles that distinguish these methods and their performance remain unclear. Here, we show that the relative performance of these methods is contingent on their ability to account for variation between biological replicates. Methods that ignore this inevitable variation are biased and prone to false discoveries. Indeed, the most widely used methods can discover hundreds of differentially expressed genes in the absence of biological differences. To exemplify these principles, we exposed true and false discoveries of differentially expressed genes in the injured mouse spinal cord.

Download Full-text

86 Co-detection of RNA and protein in FFPE tumor samples by combining RNAscope in situ hybridization and immunohistochemistry assays

Journal for ImmunoTherapy of Cancer ◽

10.1136/jitc-2020-sitc2020.0086 ◽

2020 ◽

Vol 8 (Suppl 3) ◽

pp. A95-A95

Author(s):

Anushka Dikshit ◽

Xiao-jun Ma ◽

Emerald Doolittle ◽

Lydia Hernandez ◽

Jyoti Sheldon ◽

...

Keyword(s):

Gene Expression ◽

T Cells ◽

In Situ Hybridization ◽

Single Cell ◽

Immune Cells ◽

Cell Type ◽

Ffpe Tumor ◽

Cell Type Specific ◽

Co Detection

BackgroundSpatially resolved gene expression has emerged as a crucial technique to understand complex multicellular interactions within the tumor and its microenvironment. Interrogation of complex cellular interactions within the tumor microenvironment (TME) requires a multi-omics approach where multiple RNA and protein targets can be visualized within the same tumor sample and be feasible in FFPE sample types. Simultaneous detection of RNA and protein can reveal cellular sources of secreted proteins, identify specific cell types, and visualize the spatial organization of cells within the tissue. Examination of RNA by in situ hybridization (ISH) and protein by immunohistochemistry (IHC) or immunofluorescence (IF) are widely used and accepted techniques for the detection of biomarkers in tumor samples. Given the similarities in workflow, co-detection of RNA and protein by combining ISH and IHC/IF in a single assay can be a powerful multi-omics solution for interrogating the complex tumor and its microenvironment.MethodsIn this report we combined the single cell, single molecule RNA ISH technology known as RNAscope with IHC/IF to simultaneously detect RNA and protein in the same FFPE tumor section using both chromogenic and fluorescence detection methods.ResultsWe demonstrate co-localization of target mRNA and the corresponding protein in human cancer samples, visualize infiltration of immune cells into the TME, characterize the activation state of immune cells in the TME, identify single cell gene expression within cellular boundaries demarcated by IHC/IF, examine cell type-specific expression of multiple immune checkpoint markers, and distinguish endogenous T cells from activated CAR+ T cells. Overall, we show that co-detection of RNA by the RNAscope ISH assay and protein by the IHC/IF assay in the same FFPE section is a feasible methodology. The combined RNAscope ISH-IHC/IF workflow is a powerful technique that can be used to study gene expression signatures at the RNA and protein level with spatial and single cell resolution.ConclusionsBy leveraging the strength of the similar workflows of RNAscope ISH and IHC/IF assays, this methodology combines transcriptomics and proteomics in the same tissue section, providing a multi-omics approach for characterizing complex tissues and revealing cell type specific gene expression with spatial and single cell resolution.

Download Full-text

A Markov random field model for network-based differential expression analysis of single-cell RNA-seq data

BMC Bioinformatics ◽

10.1186/s12859-021-04412-0 ◽

2021 ◽

Vol 22 (1) ◽

Author(s):

Hongyu Li ◽

Biqing Zhu ◽

Zhichao Xu ◽

Taylor Adams ◽

Naftali Kaminski ◽

...

Keyword(s):

Random Field ◽

Single Cell ◽

Differentially Expressed Genes ◽

Markov Random Field ◽

Statistical Power ◽

Cell Types ◽

Differentially Expressed ◽

Cell Type ◽

Markov Random ◽

Cell Type Specific

Abstract Background Recent development of single cell sequencing technologies has made it possible to identify genes with different expression (DE) levels at the cell type level between different groups of samples. In this article, we propose to borrow information through known biological networks to increase statistical power to identify differentially expressed genes (DEGs). Results We develop MRFscRNAseq, which is based on a Markov random field (MRF) model to appropriately accommodate gene network information as well as dependencies among cell types to identify cell-type specific DEGs. We implement an Expectation-Maximization (EM) algorithm with mean field-like approximation to estimate model parameters and a Gibbs sampler to infer DE status. Simulation study shows that our method has better power to detect cell-type specific DEGs than conventional methods while appropriately controlling type I error rate. The usefulness of our method is demonstrated through its application to study the pathogenesis and biological processes of idiopathic pulmonary fibrosis (IPF) using a single-cell RNA-sequencing (scRNA-seq) data set, which contains 18,150 protein-coding genes across 38 cell types on lung tissues from 32 IPF patients and 28 normal controls. Conclusions The proposed MRF model is implemented in the R package MRFscRNAseq available on GitHub. By utilizing gene-gene and cell-cell networks, our method increases statistical power to detect differentially expressed genes from scRNA-seq data.

Download Full-text

Single Cell Profiling Reveals Unique CXCL13 Positive T Cell Subsets in the Tumor Microenvironment of Lymphocyte Rich Classic Hodgkin Lymphoma

Blood ◽

10.1182/blood-2020-134929 ◽

2020 ◽

Vol 136 (Supplement 1) ◽

pp. 32-33

Author(s):

Tomohiro Aoki ◽

Lauren C. Chong ◽

Katsuyoshi Takata ◽

Katy Milne ◽

Elizabeth Chavez ◽

...

Keyword(s):

T Cells ◽

T Cell ◽

B Cells ◽

Single Cell ◽

Immune Cells ◽

Research Funding ◽

Immune Cell ◽

Expression Patterns ◽

The Other ◽

Tfh Cells

Introduction: Classic Hodgkin lymphoma (CHL) features a unique crosstalk between malignant cells and different types of normal immune cells in the tumor-microenvironment (TME). On the basis of histomorphologic and immunophenotypic features of the malignant Hodgkin and Reed-Sternberg (HRS) cells and infiltrating immune cells, four histological subtypes of CHL are recognized: Nodular sclerosing (NS), Mixed cellularity, Lymphocyte-rich (LR) and Lymphocyte-depleted CHL. Recently, our group described the high abundance of various types of immunosuppressive CD4+ T cells including LAG3+ and/or CTLA4+ cells in the TME of CHL using single cell RNA sequencing (scRNAseq). However, the TME of LR-CHL has not been well characterized due to the rarity of the disease. In this study, we aimed at characterizing the immune cell profile of LR-CHL at single cell resolution. METHODS: We performed scRNAseq on cell suspensions collected from lymph nodes of 28 primary CHL patients, including 11 NS, 9 MC and 8 LR samples, with 5 reactive lymph nodes (RLN) serving as normal controls. We merged the expression data from all cells (CHL and RLN) and performed batch correction and normalization. We also performed single- and multi-color immunohistochemistry (IHC) on tissue microarray (TMA) slides from the same patients. In addition, an independent validation cohort of 31 pre-treatment LR-CHL samples assembled on a TMA, were also evaluated by IHC. Results: A total of 23 phenotypic cell clusters were identified using unsupervised clustering (PhenoGraph). We assigned each cluster to a cell type based on the expression of genes described in published transcriptome data of sorted immune cells and known canonical markers. While most immune cell phenotypes were present in all pathological subtypes, we observed a lower abundance of regulatory T cells (Tregs) in LR-CHL in comparison to the other CHL subtypes. Conversely, we found that B cells were enriched in LR-CHL when compared to the other subtypes and specifically, all four naïve B-cell clusters were quantitatively dominated by cells derived from the LR-CHL samples. T follicular helper (TFH) cells support antibody response and differentiation of B cells. Our data show the preferential enrichment of TFH in LR-CHL as compared to other CHL subtypes, but TFH cells were still less frequent compared to RLN. Of note, Chemokine C-X-C motif ligand 13 (CXCL13) was identified as the most up-regulated gene in LR compared to RLN. CXCL13, which is a ligand of C-X-C motif receptor 5 (CXCR5) is well known as a B-cell attractant via the CXCR5-CXCL13 axis. Analyzing co-expression patterns on the single cell level revealed that the majority of CXCL13+ T cells co-expressed PD-1 and ICOS, which is known as a universal TFH marker, but co-expression of CXCR5, another common TFH marker, was variable. Notably, classical TFH cells co-expressing CXCR5 and PD-1 were significantly enriched in RLN, whereas PD-1+ CXCL13+ CXCR5- CD4+ T cells were significantly enriched in LR-CHL. These co-expression patterns were validated using flow cytometry. Moreover, the expression of CXCR5 on naïve B cells in the TME was increased in LR-CHL compared to the other CHL subtypes We next sought to understand the spatial relationship between CXCL13+ T cells and malignant HRS cells. IHC of all cases revealed that CXCL13+ T cells were significantly enriched in the LR-CHL TME compared to other subtypes of CHL, and 46% of the LR-CHL cases showed CXCL13+ T cell rosettes closely surrounding HRS cells. Since PD-1+ T cell rosettes are known as a specific feature of LR-CHL, we confirmed co-expression of PD-1 in the rosetting cells by IHC in these cases. Conclusions: Our results reveal a unique TME composition in LR-CHL. LR-CHL seems to be distinctly characterized among the CHL subtypes by enrichment of CXCR5+ naïve B cells and CD4+ CXCL13+ PD-1+ T cells, indicating the importance of the CXCR5-CXCL13 axis in the pathogenesis of LR-CHL. Figure Disclosures Savage: BeiGene: Other: Steering Committee; Merck, BMS, Seattle Genetics, Gilead, AstraZeneca, AbbVie: Honoraria; Roche (institutional): Research Funding; Merck, BMS, Seattle Genetics, Gilead, AstraZeneca, AbbVie, Servier: Consultancy. Scott:Janssen: Consultancy, Research Funding; Celgene: Consultancy; NanoString: Patents & Royalties: Named inventor on a patent licensed to NanoString, Research Funding; NIH: Consultancy, Other: Co-inventor on a patent related to the MCL35 assay filed at the National Institutes of Health, United States of America.; Roche/Genentech: Research Funding; Abbvie: Consultancy; AstraZeneca: Consultancy. Steidl:AbbVie: Consultancy; Roche: Consultancy; Curis Inc: Consultancy; Juno Therapeutics: Consultancy; Bayer: Consultancy; Seattle Genetics: Consultancy; Bristol-Myers Squibb: Research Funding.

Download Full-text

A new family of markers to characterize the human γδ T-cell lineage

10.31219/osf.io/jzcf3 ◽

2019 ◽

Author(s):

Shahan Mamoor

Keyword(s):

T Cells ◽

T Cell ◽

Differentially Expressed Genes ◽

Cell Lineage ◽

Γδ T Cells ◽

Differentially Expressed ◽

Γδ T Cell ◽

Differentiation Markers ◽

Hematopoietic Stem ◽

New Family

Prospective isolation of γδ T lymphocytes demands a comprehensive description of the molecules that distinguish T cells with γδ T-cell receptors (TCRs) (γδ T cells, or Tγδ) from those with αβTCRs (Tαβ). Here I describe some of the most differentially expressed genes in the γδ T cell when compared to the developmentally proximal but lineage-distinct Tαβ CD4+ and CD8+ lymphocytes. These genes encode cluster of differentiation markers, transcription factors, cell surface receptors and non-coding RNAs. As hematopoietic stem cells (HSCs) have been prospectively isolated based on the analysis of differentially expressed genes (1), any combination of these molecules may potentially be used to isolate Tγδ, perhaps even independent of the γδTCR. This description of the most striking identifying features of the Tγδ will be a resource for the isolation of a multi-potent common γδ T-cell progenitor.

Download Full-text

Global immune characterization of HBV/HCV-related hepatocellular carcinoma identifies macrophage and T-cell subsets associated with disease progression

Cell Discovery ◽

10.1038/s41421-020-00214-5 ◽

2020 ◽

Vol 6 (1) ◽

Author(s):

Guohe Song ◽

Yang Shi ◽

Meiying Zhang ◽

Shyamal Goswami ◽

Saifullah Afridi ◽

...

Keyword(s):

Hepatocellular Carcinoma ◽

T Cells ◽

T Cell ◽

Single Cell ◽

Immune Cells ◽

Immune Cell ◽

T Cell Subsets ◽

M2 Macrophage ◽

Advanced Hcc ◽

Immune Cell Subsets

AbstractDiverse immune cells in the tumor microenvironment form a complex ecosystem, but our knowledge of their heterogeneity and dynamics within hepatocellular carcinoma (HCC) still remains limited. To assess the plasticity and phenotypes of immune cells within HBV/HCV-related HCC microenvironment at single-cell level, we performed single-cell RNA sequencing on 41,698 immune cells from seven pairs of HBV/HCV-related HCC tumors and non-tumor liver tissues. We combined bio-informatic analyses, flow cytometry, and multiplex immunohistochemistry to assess the heterogeneity of different immune cell subsets in functional characteristics, transcriptional regulation, phenotypic switching, and interactions. We identified 29 immune cell subsets of myeloid cells, NK cells, and lymphocytes with unique transcriptomic profiles in HCC. A highly complex immunological network was shaped by diverse immune cell subsets that can transit among different states and mutually interact. Notably, we identified a subset of M2 macrophage with high expression of CCL18 and transcription factor CREM that was enriched in advanced HCC patients, and potentially participated in tumor progression. We also detected a new subset of activated CD8+ T cells highly expressing XCL1 that correlated with better patient survival rates. Meanwhile, distinct transcriptomic signatures, cytotoxic phenotypes, and evolution trajectory of effector CD8+ T cells from early-stage to advanced HCC were also identified. Our study provides insight into the immune microenvironment in HBV/HCV-related HCC and highlights novel macrophage and T-cell subsets that could be further exploited in future immunotherapy.

Download Full-text

Cell Type-Specific Differentiation Between Indica and Japonica Rice Root Tip Responses to Different Environments Based on Single-Cell RNA Sequencing

Frontiers in Genetics ◽

10.3389/fgene.2021.659500 ◽

2021 ◽

Vol 12 ◽

Author(s):

Zhe Wang ◽

Daofu Cheng ◽

Chengang Fan ◽

Cong Zhang ◽

Chao Zhang ◽

...

Keyword(s):

Rna Sequencing ◽

Differentially Expressed Genes ◽

Rice Root ◽

Differentially Expressed ◽

Bzip Transcription Factor ◽

Quiescent Center ◽

Cell Type ◽

Rice Varieties ◽

Specific Differentiation ◽

Cell Type Specific

Background: As Oryza sativa ssp. indica and Oryza sativa ssp. japonica are the two major subspecies of Asian cultivated rice, the adaptative evolution of these varieties in divergent environments is an important topic in both theoretical and practical studies. However, the cell type-specific differentiation between indica and japonica rice varieties in response to divergent habitat environments, which facilitates an understanding of the genetic basis underlying differentiation and environmental adaptation between rice subspecies at the cellular level, is little known.Methods: We analyzed a published single-cell RNA sequencing dataset to explore the differentially expressed genes between indica and japonica rice varieties in each cell type. To estimate the relationship between cell type-specific differentiation and environmental adaptation, we focused on genes in the WRKY, NAC, and BZIP transcription factor families, which are closely related to abiotic stress responses. In addition, we integrated five bulk RNA sequencing datasets obtained under conditions of abiotic stress, including cold, drought and salinity, in this study. Furthermore, we analyzed quiescent center cells in rice root tips based on orthologous markers in Arabidopsis.Results: We found differentially expressed genes between indica and japonica rice varieties with cell type-specific patterns, which were enriched in the pathways related to root development and stress reposes. Some of these genes were members of the WRKY, NAC, and BZIP transcription factor families and were differentially expressed under cold, drought or salinity stress. In addition, LOC_Os01g16810, LOC_Os01g18670, LOC_Os04g52960, and LOC_Os08g09350 may be potential markers of quiescent center cells in rice root tips.Conclusion: These results identified cell type-specific differentially expressed genes between indica-japonica rice varieties that were related to various environmental stresses and provided putative markers of quiescent center cells. This study provides new clues for understanding the development and physiology of plants during the process of adaptative divergence, in addition to identifying potential target genes for the improvement of stress tolerance in rice breeding applications.

Download Full-text

Systematic and Cell Type-Specific Telomere Length Changes in Subsets of Lymphocytes

Journal of Immunology Research ◽

10.1155/2016/5371050 ◽

2016 ◽

Vol 2016 ◽

pp. 1-9 ◽

Cited By ~ 48

Author(s):

Jue Lin ◽

Joshua Cheon ◽

Rashida Brown ◽

Michael Coccia ◽

Eli Puterman ◽

...

Keyword(s):

T Cells ◽

T Cell ◽

B Cells ◽

Telomere Length ◽

Cell Types ◽

Cell Type ◽

Peripheral Blood Mononuclear ◽

Potential Biomarker ◽

Cell Type Specific ◽

Length Changes

Telomeres, the protective DNA-protein complexes at the ends of linear chromosomes, are important for genome stability. Leukocyte or peripheral blood mononuclear cell (PBMC) telomere length is a potential biomarker for human aging that integrates genetic, environmental, and lifestyle factors and is associated with mortality and risks for major diseases. However, only a limited number of studies have examined longitudinal changes of telomere length and few have reported data on sorted circulating immune cells. We examined the average telomere length (TL) in CD4+, CD8+CD28+, and CD8+CD28− T cells, B cells, and PBMCs, cross-sectionally and longitudinally, in a cohort of premenopausal women. We report that TL changes over 18 months were correlated among these three T cell types within the same participant. Additionally, PBMC TL change was also correlated with those of all three T cell types, and B cells. The rate of shortening for B cells was significantly greater than for the three T cell types. CD8+CD28− cells, despite having the shortest TL, showed significantly more rapid attrition when compared to CD8+CD28+ T cells. These results suggest systematically coordinated, yet cell type-specific responses to factors and pathways contribute to telomere length regulation.

Download Full-text

scConsensus: combining supervised and unsupervised clustering for cell type identification in single-cell RNA sequencing data

BMC Bioinformatics ◽

10.1186/s12859-021-04028-4 ◽

2021 ◽

Vol 22 (1) ◽

Author(s):

Bobby Ranjan ◽

Florian Schmidt ◽

Wenjie Sun ◽

Jinyu Park ◽

Mohammad Amin Honardoost ◽

...

Keyword(s):

Single Cell ◽

Rna Sequencing ◽

Differentially Expressed Genes ◽

Cell Types ◽

Unsupervised Clustering ◽

Differentially Expressed ◽

Consensus Clustering ◽

Cell Type ◽

Sequencing Data ◽

Single Cell Rna Sequencing

Abstract Background Clustering is a crucial step in the analysis of single-cell data. Clusters identified in an unsupervised manner are typically annotated to cell types based on differentially expressed genes. In contrast, supervised methods use a reference panel of labelled transcriptomes to guide both clustering and cell type identification. Supervised and unsupervised clustering approaches have their distinct advantages and limitations. Therefore, they can lead to different but often complementary clustering results. Hence, a consensus approach leveraging the merits of both clustering paradigms could result in a more accurate clustering and a more precise cell type annotation. Results We present scConsensus, an $${\mathbf {R}}$$ R framework for generating a consensus clustering by (1) integrating results from both unsupervised and supervised approaches and (2) refining the consensus clusters using differentially expressed genes. The value of our approach is demonstrated on several existing single-cell RNA sequencing datasets, including data from sorted PBMC sub-populations. Conclusions scConsensus combines the merits of unsupervised and supervised approaches to partition cells with better cluster separation and homogeneity, thereby increasing our confidence in detecting distinct cell types. scConsensus is implemented in $${\mathbf {R}}$$ R and is freely available on GitHub at https://github.com/prabhakarlab/scConsensus.

Download Full-text