Get the Cytobank Platform and Publish

Single-cell analysis using algorithms: how and why?

For a general introduction on why to use machine learning-assisted analysis for cytometry data visit our page here.

In the following papers you will find an overview of analytical approaches for CyTOF data as well as a description of tools in the Cytobank platform such as viSNE, SPADE and CITRUS.

  • J Immunol. 2018 Jan 1; 200(1): 3–22. doi: 10.4049/jimmunol.1701494. A Beginner’s Guide To Analyzing and Visualizing Mass Cytometry Data. 

  • J Invest Dermatol. 2017 May;137(5):e43-e51. doi: 10.1016/j.jid.2017.03.002. Research Techniques Made Simple: Mass Cytometry Analysis Tools for Decrypting the Complexity of Biological Systems. 

Clustering algorithms, biomarkers discovery and dimensionality reduction: when ML-assisted analysis supports outstanding immunology research

CITRUS (cluster identification, characterization, and regression) is an algorithm designed for the fully automated discovery of statistically significant stratifying biological signatures within single cell datasets containing numerous samples across multiple known endpoints (e.g., responders versus non-responders).

In the following papers you will find application of CITRUS in basic and translational research.

  • Cell Rep. 2020 Sep 29;32(13):108204. doi: 10.1016/j.celrep.2020.108204. Clusters of Tolerogenic B Cells Feature in the Dynamic Immunological Landscape of the Pregnant Uterus
    Aim: Assess the uterine immune signature before and during pregnancy.
    Approach: Flow cytometry staining on whole blood or isolated mononuclear cell populations
    Advanced algorithms in Cytobank: CITRUS
    Results: CITRUS was used to identify clusters of B cells present in decidua but not in peripheral blood.

  • Cell Rep. 2019 Feb 19;26(8):2178-2193.e3. doi: 10.1016/j.celrep.2019.01.085. Mass Cytometry Analysis Reveals that Specific Intratumoral CD4+T Cell Subsets Correlate with Patient Survival in Follicular Lymphoma
    Aim: Identify biomarkers in intratumoral T cell infiltrates associated with patient survival in follicular lymphoma.
    Approach: Phenotypic single cell analysis of tumor samples by CyTOF
    Advanced algorithms in Cytobank: viSNE, SPADE, CITRUS 
    Results: Identified PD-1+ T cell clusters associated with poor survival, while traditional analysis of # of PD-1+ T cells was insufficient.

  • Front Immunol. 2018 Dec 5;9:2783. doi: 10.3389/fimmu.2018.02783. eCollection 2018. Multivariate Computational Analysis of Gamma Delta T Cell Inhibitory Receptor Signatures Reveals the Divergence of Healthy and ART-Suppressed HIV+ Aging.
    Aim: Identify immune signatures predictive of anti-retroviral therapy-suppressed aging in HIV infection.
    Approach: Flow cytometry of PBMCs
    Advanced algorithms in Cytobank: CITRUS 
    Results: Identified gamma delta T cell inhibitory receptor expression signature associated with ART-suppressed HIV+ subjects compared to healthy controls.

  • J Autoimmun. 2017 Apr 4. pii: S0896-8411(16)30412-7. doi: 10.1016/j.jaut.2017.03.010. Mass cytometry identifies a distinct monocyte cytokine signature shared by clinically heterogeneous pediatric SLE patients
    Aim: Identify biomarkers in peripheral whole blood associated with severe systemic lupus erythematosus in pediatric patients versus healthy controls.
    Approach: Phenotypic and functional single cell analysis of peripheral blood by CyTOF
    Advanced algorithms in Cytobank: CITRUS, viSNE 
    Results: Identified clusters of activated monocytes predictive of pediatric SLE using CITRUS.

viSNE is an algorithm that reduces high-parameter data down to two dimensions for easy visualization and rapid exploratory data analysis of any data type. viSNE in Cytobank uses the Barnes-Hut implementation of the t-SNE algorithm.

FlowSOM is an algorithm that speeds time to analysis and quality of clustering with Self-Organizing Maps (SOMs) that can reveal how all markers are behaving on all cells, and can detect subsets that might otherwise be missed. It clusters cells (or other observations) based on chosen clustering channels (or markers/features), generates a SOM of clusters, produces a Minimum Spanning Tree (MST) of the clusters, and assigns each cluster to a metacluster, effectively grouping them into a population.

In the following papers you will find application of a viSNE and FlowSOM workflow to answer common research questions.

  • Science. 2020 Sep 4;369(6508):eabc8511. doi: 10.1126/science.abc8511. Epub 2020 Jul 15. Deep immune profiling of COVID-19 patients reveals distinct immunotypes with therapeutic implications
    Aim: Immune profiling of COVID-19 patients.
    Approach: High dimensional cytometry on PBMC from healthy donors and recovered or current COVID-19 patients
    Advanced algorithms in Cytobank: viSNE, FlowSOM.
    Results: Identified clusters had different abundancies among the groups of patients.
  • Cell Host Microbe. 2020 Aug 12; S1931-3128(20)30410-8. doi: 10.1016/j.chom.2020.07.014 Programming of an Intravascular Immune Firewall by the Gut Microbiota Protects against Pathogen Dissemination during Infection
    Aim: Determine the functional role of the gut microbiota in host defense against bloodstream infections by the intravascular firewall formed by KCs in the liver.
    Approach: Mass cytometry analysis of liver non-parenchymal cells.
    Advanced algorithms in Cytobank: viSNE, FlowSOM.
    Results: Functional differences in bacterial capture by KCs in GF versus colonized mice could not be explained by microbiota-dependent shifts in KC subsets identified by FlowSOM clusters.

  • Science. 2020 Aug 7;369(6504):718-724. doi: 10.1126/science.abc6027. Epub 2020 Jul 13. Impaired type I interferon activity and inflammatory responses in severe COVID-19 patients
    Aim: Perform an integrated immune analysis of a cohort of 50 COVID-19 patients 
    Approach: Multiparameter phenotyping of peripheral blood leukocytes using mass cytometry 
    Advanced algorithms in Cytobank: viSNE
    Results: Severe and critical patients had a highly impaired IFN type I response as well as a persistent blood viral load and exacerbated inflammatory response     

SPADE stands for Spanning-tree Progression Analysis of Density-normalized Events. SPADE clusters phenotypically similar cells into a hierarchy that allows high-throughput, multidimensional analysis of heterogeneous samples.

In the following papers you will find application of SPADE in basic and translational research.

  • Nature. 2020 Oct;586(7827):101-107. doi: 10.1038/s41586-020-2734-6. Epub 2020 Sep 16. Reprogramming roadmap reveals route to human induced trophoblast stem cells
    Aim: Identify molecular mechanisms underlying the reprogramming of human somatic cells to primed or naïve induced pluripotent stem cells. 
    Approach: Single cell transcriptomics to reconstruct molecular reprogramming trajectories of human fibroblasts 
    Advanced algorithms in Cytobank: SPADE 
    Results: SPADE trees helped in visualizing flow cytometry profile intermediates during reprogramming.

  • J Exp Med. 2019 Sep 2;216(9):2071-2090. doi: 10.1084/jem.20181124. Epub 2019 Jun 20. Widespread B cell perturbations in HIV-1 infection afflict naive and marginal zone B cells
    Aim: Deeply analyze B cell perturbations in HIV-1 infection. 
    Approach: B cell specific high-dimensional flow cytometry on cryopreserved peripheral blood mononuclear cells from healthy donors and HIV-1 infected patients
    Advanced algorithms in Cytobank: SPADE 
    Results: SPADE trees revealed heterogeneity of CD21neg naive B cells in chronic HIV-1 infection

  • Front Immunol. 2018 Aug 3;9:1613. doi: 10.3389/fimmu.2018.01613. eCollection 2018. Circulating T Cell Subpopulations Correlate With Immune Responses at the Tumor Site and Clinical Response to PD1 Inhibition in Non-Small Cell Lung Cancer. 
    Aim: Identify peripheral blood correlates of anti-tumor response in absence of checkpoint inhibitors in melanoma, non-small cell lung cancer.
    Approach: T-cell focused flow cytometry and gene expression on whole peripheral blood 
    Advanced algorithms in Cytobank: SPADE 
    Results: SPADE illustrated differing T cell differentiation patterns in melanoma patients. Proposed these subpopulations could be used to predict response to checkpoint inhibitors.
  • Clin Transl Immunology. 2020 May 5;9(5):e1127. doi: 10.1002/cti2.1127. eCollection 2020 May. High-dimensional analyses reveal a distinct role of T-cell subsets in the immune microenvironment of gastric cancer
    Aim: Immune cellular and molecular analyses on tumor tissues and peripheral blood of gastric cancer patients.
    Approach: Mass cytometry (CyTOF) in peripheral blood from selected patients.
    Advanced algorithms in Cytobank: viSNE, SPADE
    Results: Identified clusters were differentially expressed in groups with good or poor outcomes
  • Clin Transl Immunology. 2020 Apr 29;9(5):e01132. doi: 10.1002/cti2.1132. eCollection 2020 May Identification of the immune checkpoint signature of multiple myeloma using mass cytometry-based single-cell analysis
    Aim: Overview of the different immune subsets in the MM microenvironment.
    Approach: Mass cytometry on bone marrow samples of MM patients
    Advanced algorithms in Cytobank: viSNE, SPADE
    Results: Identified clusters with different median expression of immune checkpoint protein were specifically present in MM patients