Critical assessment of automated flow cytometry data analysis techniques

IMMUNOLOGY. An interactive reference framework for modeling a dynamic immune system

Immune cells function in an interacting hierarchy that coordinates the activities of various cell types according to genetic and environmental contexts. We developed graphical approaches to construct an extensible immune reference map from mass cytometry data of cells from different organs, incorporating landmark cell populations as flags on the map to compare cells from distinct samples. The maps recapitulated canonical cellular phenotypes and revealed reproducible, tissue-specific deviations. The approach revealed influences of genetic variation and circadian rhythms on immune system structure, enabled direct comparisons of murine and human blood cell phenotypes, and even enabled archival fluorescence-based flow cytometry data to be mapped onto the reference framework. This foundational reference map provides a working definition of systemic immune organization to which new data can be integrated to reveal deviations driven by genetics, environment, or pathology.

Geometry of the Gene Expression Space of Individual Cells

There is a revolution in the ability to analyze gene expression of single cells in a tissue. To understand this data we must comprehend how cells are distributed in a high-dimensional gene expression space. One open question is whether cell types form discrete clusters or whether gene expression forms a continuum of states. If such a continuum exists, what is its geometry? Recent theory on evolutionary trade-offs suggests that cells that need to perform multiple tasks are arranged in a polygon or polyhedron (line, triangle, tetrahedron and so on, generally called polytopes) in gene expression space, whose vertices are the expression profiles optimal for each task. Here, we analyze single-cell data from human and mouse tissues profiled using a variety of single-cell technologies. We fit the data to shapes with different numbers of vertices, compute their statistical significance, and infer their tasks. We find cases in which single cells fill out a continuum of expression states within a polyhedron. This occurs in intestinal progenitor cells, which fill out a tetrahedron in gene expression space. The four vertices of this tetrahedron are each enriched with genes for a specific task related to stemness and early differentiation. A polyhedral continuum of states is also found in spleen dendritic cells, known to perform multiple immune tasks: cells fill out a tetrahedron whose vertices correspond to key tasks related to maturation, pathogen sensing and communication with lymphocytes. A mixture of continuum-like distributions and discrete clusters is found in other cell types, including bone marrow and differentiated intestinal crypt cells. This approach can be used to understand the geometry and biological tasks of a wide range of single-cell datasets. The present results suggest that the concept of cell type may be expanded. In addition to discreet clusters in gene-expression space, we suggest a new possibility: a continuum of states within a polyhedron, in which the vertices represent specialists at key tasks.

In the past, biological experiments usually pooled together millions of cells, masking the differences between individual cells. Current technology takes a big step forward by measuring gene expression from individual cells. Interpreting this data is challenging because we need to understand how cells are arranged in a high dimensional gene expression space. Here we test recent theory that suggests that cells facing multiple tasks should be arranged in simple low dimensional polygons or polyhedra (generally called polytopes). The vertices of the polytopes are gene expression profiles optimal for each of the tasks. We find evidence for such simplicity in a variety of tissues—spleen, bone marrow, intestine—analyzed by different single-cell technologies. We find that cells are distributed inside polytopes, such as tetrahedrons or four-dimensional simplexes, with cells closest to each vertex responsible for a different key task. For example, intestinal progenitor cells that give rise to the other cell types show a continuous distribution in a tetrahedron whose vertices correspond to several key sub-tasks. Immune dendritic cells likewise are continuously distributed between key immune tasks. This approach of testing whether data falls in polytopes may be useful for interpreting a variety of single-cell datasets in terms of biological tasks.

Computational prediction of manually gated rare cells in flow cytometry data

Rare cell identification is an interesting and challenging question in flow cytometry data analysis. In the literature, manual gating is a popular approach to distill flow cytometry data and drill down to the rare cells of interest, based on prior knowledge of measured protein markers and visual inspection of the data. Several computational algorithms have been proposed for rare cell identification. To compare existing algorithms and promote new developments, FlowCAP-III put forward one computational challenge that focused on this question. The challenge provided flow cytometry data for 202 training samples and two manually gated rare cell types for each training sample, roughly 0.02 and 0.04% of the cells, respectively. In addition, flow cytometry data for 203 testing samples were provided, and participants were invited to computationally identify the rare cells in the testing samples. Accuracy of the identification results was evaluated by comparing to manual gating of the testing samples. We participated in the challenge, and developed a method that combined the Hellinger divergence, a downsampling trick and the ensemble SVM. Our method achieved the highest accuracy in the challenge.

Identification and visualization of multidimensional antigen-specific T-cell populations in polychromatic cytometry data

An important aspect of immune monitoring for vaccine development, clinical trials, and research is the detection, measurement, and comparison of antigen-specific T-cells from subject samples under different conditions. Antigen-specific T-cells compose a very small fraction of total T-cells. Developments in cytometry technology over the past five years have enabled the measurement of single-cells in a multivariate and high-throughput manner. This growth in both dimensionality and quantity of data continues to pose a challenge for effective identification and visualization of rare cell subsets, such as antigen-specific T-cells. Dimension reduction and feature extraction play pivotal role in both identifying and visualizing cell populations of interest in large, multi-dimensional cytometry datasets. However, the automated identification and visualization of rare, high-dimensional cell subsets remains challenging. Here we demonstrate how a systematic and integrated approach combining targeted feature extraction with dimension reduction can be used to identify and visualize biological differences in rare, antigen-specific cell populations. By using OpenCyto to perform semi-automated gating and features extraction of flow cytometry data, followed by dimensionality reduction with t-SNE we are able to identify polyfunctional subpopulations of antigen-specific T-cells and visualize treatment-specific differences between them.

Methods for discovery and characterization of cell subsets in high dimensional mass cytometry data

The flood of high-dimensional data resulting from mass cytometry experiments that measure more than 40 features of individual cells has stimulated creation of new single cell computational biology tools. These tools draw on advances in the field of machine learning to capture multi-parametric relationships and reveal cells that are easily overlooked in traditional analysis. Here, we introduce a workflow for high dimensional mass cytometry data that emphasizes unsupervised approaches and visualizes data in both single cell and population level views. This workflow includes three central components that are common across mass cytometry analysis approaches: (1) distinguishing initial populations, (2) revealing cell subsets, and (3) characterizing subset features. In the implementation described here, viSNE, SPADE, and heatmaps were used sequentially to comprehensively characterize and compare healthy and malignant human tissue samples. The use of multiple methods helps provide a comprehensive view of results, and the largely unsupervised workflow facilitates automation and helps researchers avoid missing cell populations with unusual or unexpected phenotypes. Together, these methods develop a framework for future machine learning of cell identity.

Data-Driven Phenotypic Dissection of AML Reveals Progenitor-like Cells that Correlate with Prognosis

Acute myeloid leukemia (AML) manifests as phenotypically and functionally diverse cells, often within the same patient. Intratumor phenotypic and functional heterogeneity have been linked primarily by physical sorting experiments, which assume that functionally distinct subpopulations can be prospectively isolated by surface phenotypes. This assumption has proven problematic, and we therefore developed a data-driven approach. Using mass cytometry, we profiled surface and intracellular signaling proteins simultaneously in millions of healthy and leukemic cells. We developed PhenoGraph, which algorithmically defines phenotypes in high-dimensional single-cell data. PhenoGraph revealed that the surface phenotypes of leukemic blasts do not necessarily reflect their intracellular state. Using hematopoietic progenitors, we defined a signaling-based measure of cellular phenotype, which led to isolation of a gene expression signature that was predictive of survival in independent cohorts. This study presents new methods for large-scale analysis of single-cell heterogeneity and demonstrates their utility, yielding insights into AML pathophysiology.

Computationally efficient multidimensional analysis of complex flow cytometry data using second order polynomial histograms

Many methods have been described for automated clustering analysis of complex flow cytometry data, but so far the goal to efficiently estimate multivariate densities and their modes for a moderate number of dimensions and potentially millions of data points has not been attained. We have devised a novel approach to describing modes using second order polynomial histogram estimators (SOPHE). The method divides the data into multivariate bins and determines the shape of the data in each bin based on second order polynomials, which is an efficient computation. These calculations yield local maxima and allow joining of adjacent bins to identify clusters. The use of second order polynomials also optimally uses wide bins, such that in most cases each parameter (dimension) need only be divided into 4-8 bins, again reducing computational load. We have validated this method using defined mixtures of up to 17 fluorescent beads in 16 dimensions, correctly identifying all populations in data files of 100,000 beads in <10 s, on a standard laptop. The method also correctly clustered granulocytes, lymphocytes, including standard T, B, and NK cell subsets, and monocytes in 9-color stained peripheral blood, within seconds. SOPHE successfully clustered up to 36 subsets of memory CD4 T cells using differentiation and trafficking markers, in 14-color flow analysis, and up to 65 subpopulations of PBMC in 33-dimensional CyTOF data, showing its usefulness in discovery research. SOPHE has the potential to greatly increase efficiency of analysing complex mixtures of cells in higher dimensions.

Managing Multi-center Flow Cytometry Data for Immune Monitoring

With the recent results of promising cancer vaccines and immunotherapy1–5, immune monitoring has become increasingly relevant for measuring treatment-induced effects on T cells, and an essential tool for shedding light on the mechanisms responsible for a successful treatment. Flow cytometry is the canonical multi-parameter assay for the fine characterization of single cells in solution, and is ubiquitously used in pre-clinical tumor immunology and in cancer immunotherapy trials. Current state-of-the-art polychromatic flow cytometry involves multi-step, multi-reagent assays followed by sample acquisition on sophisticated instruments capable of capturing up to 20 parameters per cell at a rate of tens of thousands of cells per second. Given the complexity of flow cytometry assays, reproducibility is a major concern, especially for multi-center studies. A promising approach for improving reproducibility is the use of automated analysis borrowing from statistics, machine learning and information visualization21–23, as these methods directly address the subjectivity, operator-dependence, labor-intensive and low fidelity of manual analysis. However, it is quite time-consuming to investigate and test new automated analysis techniques on large data sets without some centralized information management system. For large-scale automated analysis to be practical, the presence of consistent and high-quality data linked to the raw FCS files is indispensable. In particular, the use of machine-readable standard vocabularies to characterize channel metadata is essential when constructing analytic pipelines to avoid errors in processing, analysis and interpretation of results. For automation, this high-quality metadata needs to be programmatically accessible, implying the need for a consistent Application Programming Interface (API). In this manuscript, we propose that upfront time spent normalizing flow cytometry data to conform to carefully designed data models enables automated analysis, potentially saving time in the long run. The ReFlow informatics framework was developed to address these data management challenges.

Consensus guidelines on plasma cell myeloma minimal residual disease analysis and reporting

BACKGROUND

Major heterogeneity between laboratories in flow cytometry (FC) minimal residual disease (MRD) testing in multiple myeloma (MM) must be overcome. Cytometry societies such as the International Clinical Cytometry Society and the European Society for Clinical Cell Analysis recognize a strong need to establish minimally acceptable requirements and recommendations to perform such complex testing.

METHODS

A group of 11 flow cytometrists currently performing FC testing in MM using different instrumentation, panel designs (≥ 6-color) and analysis software compared the procedures between their respective laboratories and reviewed the literature to propose a consensus guideline on flow-MRD analysis and reporting in MM.

RESULTS/CONCLUSION

Consensus guidelines support i) the use of minimum of five initial gating parameters (CD38, CD138, CD45, forward, and sideward light scatter) within the same aliquot for accurate identification of the total plasma cell compartment; ii) the analysis of potentially aberrant phenotypic markers and to report the antigen expression pattern on neoplastic plasma cells as being reduced, normal or increased, when compared to a normal reference plasma cell immunophenotype (obtained using the same instrument and parameters); and iii) the percentage of total bone marrow plasma cells plus the percentages of both normal and neoplastic plasma cells within the total bone marrow plasma cell compartment, and over total bone marrow cells. Consensus guidelines on minimal current and future MRD analyses should target a lower limit of detection of 0.001%, and ideally a limit of quantification of 0.001%, which requires at least 3 × 10(6) and 5 × 10(6) bone marrow cells to be measured, respectively.

COMPASS identifies T-cell subsets correlated with clinical outcomes

Advances in flow cytometry and other single-cell technologies have enabled high-dimensional, high-throughput measurements of individual cells as well as the interrogation of cell population heterogeneity. However, in many instances, computational tools to analyze the wealth of data generated by these technologies are lacking. Here, we present a computational framework for unbiased combinatorial polyfunctionality analysis of antigen-specific T-cell subsets (COMPASS). COMPASS uses a Bayesian hierarchical framework to model all observed cell subsets and select those most likely to have antigen-specific responses. Cell-subset responses are quantified by posterior probabilities, and human subject-level responses are quantified by two summary statistics that describe the quality of an individual's polyfunctional response and can be correlated directly with clinical outcome. Using three clinical data sets of cytokine production, we demonstrate how COMPASS improves characterization of antigen-specific T cells and reveals cellular 'correlates of protection/immunity' in the RV144 HIV vaccine efficacy trial that are missed by other methods. COMPASS is available as open-source software.

Studies of Antibiotic Resistance of Beta-Lactamase Bacteria under Different Nutrition Limitations at the Single-Cell Level

Drug resistance involves many biological processes, including cell growth, cell communication, and cell cooperation. In the last few decades, bacterial drug resistance studies have made substantial progress. However, a major limitation of the traditional resistance study still exists: most of the studies have concentrated on the average behavior of enormous amounts of cells rather than surveying single cells with different phenotypes or genotypes. Here, we report our study of beta-lactamase bacterial drug resistance in a well-designed microfluidic device, which allows us to conduct more controllable experiments, such as controlling the nutrient concentration, switching the culture media, performing parallel experiments, observing single cells, and acquiring time-lapse images. By using GFP as a beta-lactamase indicator and acquiring time-lapse images at the single-cell level, we observed correlations between the bacterial heterogeneous phenotypes and their behavior in different culture media. The feedback loop between the growth rate and the beta-lactamase production suggests that the beta-lactamase bacteria are more resistant in a rich medium than in a relatively poor medium. In the poorest medium, the proportion of dormant cells may increase, which causes a lower death rate in the same generation. Our work may contribute to assaying the antibiotic resistance of pathogenic bacteria in heterogeneous complex media.

Mass cytometry as a platform for the discovery of cellular biomarkers to guide effective rheumatic disease therapy

The development of biomarkers for autoimmune diseases has been hampered by a lack of understanding of disease etiopathogenesis and of the mechanisms underlying the induction and maintenance of inflammation, which involves complex activation dynamics of diverse cell types. The heterogeneous nature and suboptimal clinical response to treatment observed in many autoimmune syndromes highlight the need to develop improved strategies to predict patient outcome to therapy and personalize patient care. Mass cytometry, using CyTOF®, is an advanced technology that facilitates multiparametric, phenotypic analysis of immune cells at single-cell resolution. In this review, we outline the capabilities of mass cytometry and illustrate the potential of this technology to enhance the discovery of cellular biomarkers for rheumatoid arthritis, a prototypical autoimmune disease.

Data analysis as a source of variability of the HLA-peptide multimer assay: from manual gating to automated recognition of cell clusters

Multiparameter flow cytometry is an indispensable method for assessing antigen-specific T cells in basic research and cancer immunotherapy. Proficiency panels have shown that cell sample processing, test protocols and data analysis may all contribute to the variability of the results obtained by laboratories performing ex vivo T cell immune monitoring. In particular, analysis currently relies on a manual, step-by-step strategy employing serial gating decisions based on visual inspection of one- or two-dimensional plots. It is therefore operator dependent and subjective. In the context of continuing efforts to support inter-laboratory T cell assay harmonization, the CIMT Immunoguiding Program organized its third proficiency panel dedicated to the detection of antigen-specific CD8(+) T cells by HLA-peptide multimer staining. We first assessed the contribution of manual data analysis to the variability of reported T cell frequencies within a group of laboratories staining and analyzing the same cell samples with their own reagents and protocols. The results show that data analysis is a source of variation in the multimer assay outcome. To evaluate whether an automated analysis approach can reduce variability of proficiency panel data, we used a hierarchical statistical mixture model to identify cell clusters. Challenges for automated analysis were the need to process non-standardized data sets from multiple centers, and the fact that the antigen-specific cell frequencies were very low in most samples. We show that this automated method can circumvent difficulties inherent to manual gating strategies and is broadly applicable for experiments performed with heterogeneous protocols and reagents.

A mine is a terrible thing to waste: high content, single cell technologies for comprehensive immune analysis

In recent years, an incredible variety of single cell technologies have become available to analyze immune responses. These technologies include polychromatic flow cytometry, mass cytometry, highly multiplexed single cell qPCR, RNA sequencing, microtools, and high-resolution imaging. In this article, we review these platforms, describing their power and limitations for comprehensive analysis of the immune system. We relate the properties of these technologies to the various cellular states relevant to an immune response, in order to address which technologies are most appropriate for which settings.

immunoClust--An automated analysis pipeline for the identification of immunophenotypic signatures in high-dimensional cytometric datasets

Multiparametric fluorescence and mass cytometry offers new perspectives to disclose and to monitor the high diversity of cell populations in the peripheral blood for biomarker research. While high-end cytometric devices are currently available to detect theoretically up to 120 individual parameters at the single cell level, software tools are needed to analyze these complex datasets automatically in acceptable time and without operator bias or knowledge. We developed an automated analysis pipeline, immunoClust, for uncompensated fluorescence and mass cytometry data, which consists of two parts. First, cell events of each sample are grouped into individual clusters. Subsequently, a classification algorithm assorts these cell event clusters into populations comparable between different samples. The clustering of cell events is designed for datasets with large event counts in high dimensions as a global unsupervised method, sensitive to identify rare cell types even when next to large populations. Both parts use model-based clustering with an iterative expectation maximization algorithm and the integrated classification likelihood to obtain the clusters. A detailed description of both algorithms is presented. Testing and validation was performed using 1) blood cell samples of defined composition that were depleted of particular cell subsets by magnetic cell sorting, 2) datasets of the FlowCAP III challenges to identify populations of rare cell types and 3) high-dimensional fluorescence and mass-cytometry datasets for comparison with conventional manual gating procedures. In conclusion, the immunoClust-algorithm is a promising tool to standardize and automate the analysis of high-dimensional cytometric datasets. As a prerequisite for interpretation of such data, it will support our efforts in developing immunological biomarkers for chronic inflammatory disorders and therapy recommendations in personalized medicine. immunoClust is implemented as an R-package and is provided as source code from www.bioconductor.org.

Polychromatic flow cytometry in evaluating rheumatic disease patients

B cells are central players in multiple autoimmune rheumatic diseases as a result of the imbalance between pathogenic and protective B-cell functions, which are presumably mediated by distinct populations. Yet the functional role of different B-cell populations and the contribution of specific subsets to disease pathogenesis remain to be fully understood owing to a large extent to the use of pauci-color flow cytometry. Despite its limitations, this approach has been instrumental in providing a global picture of multiple B-cell abnormalities in multiple human rheumatic diseases, more prominently systemic lupus erythematosus, rheumatoid arthritis and Sjogren’s syndrome. Accordingly, these studies represent the focus of this review. In addition, we also discuss the added value of tapping into the potential of polychromatic flow cytometry to unravel a higher level of B-cell heterogeneity, provide a more nuanced view of B-cell abnormalities in disease and create the foundation for a precise understanding of functional division of labor among the different phenotypic subsets. State-of-the-art polychromatic flow cytometry and novel multidimensional analytical approaches hold tremendous promise for our understanding of disease pathogenesis, the generation of disease biomarkers, patient stratification and personalized therapeutic approaches.

Complexity and challenges in defining myeloid-derived suppressor cells

Study of myeloid cells endowed with suppressive activity is an active field of research which has particular importance in cancer, in view of the negative regulatory capacity of these cells to the host's immune response. The expansion of these cells, called myeloid-derived suppressor cells (MDSCs), has been documented in many models of tumor-bearing mice and in patients with tumors of various origin, and their presence is associated with disease progression and reduced survival. For this reason, monitoring this type of cell expansion is of clinical importance, and flow cytometry is the technique of choice for their identification. Over the years, it has been demonstrated that MDSCs comprise a group of immature myeloid cells belonging both to monocytic and granulocytic lineages, with several stages of differentiation; their occurrence depends on tumor-derived soluble factors, which guide their expansion and determine their block of differentiation. Because of their heterogeneous composition, accurate phenotyping of these cells requires a multicolor approach, so that the expansion of all MDSC subsets can be appreciated. This review article focuses on identifying MDSCs and discusses problems associated with phenotyping circulating and tumor-associated MDSCs in humans and in mouse models.

Towards a commercial process for the manufacture of genetically modified T cells for therapy

The recent successes of adoptive T-cell immunotherapy for the treatment of hematologic malignancies have highlighted the need for manufacturing processes that are robust and scalable for product commercialization. Here we review some of the more outstanding issues surrounding commercial scale manufacturing of personalized-adoptive T-cell medicinal products. These include closed system operations, improving process robustness and simplifying work flows, reducing labor intensity by implementing process automation, scalability and cost, as well as appropriate testing and tracking of products, all while maintaining strict adherence to Current Good Manufacturing Practices and regulatory guidelines. A decentralized manufacturing model is proposed, where in the future patients' cells could be processed at the point-of-care in the hospital.

Deep profiling of multitube flow cytometry data

MOTIVATION

Deep profiling the phenotypic landscape of tissues using high-throughput flow cytometry (FCM) can provide important new insights into the interplay of cells in both healthy and diseased tissue. But often, especially in clinical settings, the cytometer cannot measure all the desired markers in a single aliquot. In these cases, tissue is separated into independently analysed samples, leaving a need to electronically recombine these to increase dimensionality. Nearest-neighbour (NN) based imputation fulfils this need but can produce artificial subpopulations. Clustering-based NNs can reduce these, but requires prior domain knowledge to be able to parameterize the clustering, so is unsuited to discovery settings.

RESULTS

We present flowBin, a parameterization-free method for combining multitube FCM data into a higher-dimensional form suitable for deep profiling and discovery. FlowBin allocates cells to bins defined by the common markers across tubes in a multitube experiment, then computes aggregate expression for each bin within each tube, to create a matrix of expression of all markers assayed in each tube. We show, using simulated multitube data, that flowType analysis of flowBin output reproduces the results of that same analysis on the original data for cell types of >10% abundance. We used flowBin in conjunction with classifiers to distinguish normal from cancerous cells. We used flowBin together with flowType and RchyOptimyx to profile the immunophenotypic landscape of NPM1-mutated acute myeloid leukemia, and present a series of novel cell types associated with that mutation.

High Reproducibility of ELISPOT Counts from Nine Different Laboratories

The primary goal of immune monitoring with ELISPOT is to measure the number of T cells, specific for any antigen, accurately and reproducibly between different laboratories. In ELISPOT assays, antigen-specific T cells secrete cytokines, forming spots of different sizes on a membrane with variable background intensities. Due to the subjective nature of judging maximal and minimal spot sizes, different investigators come up with different numbers. This study aims to determine whether statistics-based, automated size-gating can harmonize the number of spot counts calculated between different laboratories. We plated PBMC at four different concentrations, 24 replicates each, in an IFN-γ ELISPOT assay with HCMV pp65 antigen. The ELISPOT plate, and an image file of the plate was counted in nine different laboratories using ImmunoSpot® Analyzers by (A) Basic Count™ relying on subjective counting parameters set by the respective investigators and (B) SmartCount™, an automated counting protocol by the ImmunoSpot® Software that uses statistics-based spot size auto-gating with spot intensity auto-thresholding. The average coefficient of variation (CV) for the mean values between independent laboratories was 26.7% when counting with Basic Count™, and 6.7% when counting with SmartCount™. Our data indicates that SmartCount™ allows harmonization of counting ELISPOT results between different laboratories and investigators.

FlowSOM: Using self-organizing maps for visualization and interpretation of cytometry data

The number of markers measured in both flow and mass cytometry keeps increasing steadily. Although this provides a wealth of information, it becomes infeasible to analyze these datasets manually. When using 2D scatter plots, the number of possible plots increases exponentially with the number of markers and therefore, relevant information that is present in the data might be missed. In this article, we introduce a new visualization technique, called FlowSOM, which analyzes Flow or mass cytometry data using a Self-Organizing Map. Using a two-level clustering and star charts, our algorithm helps to obtain a clear overview of how all markers are behaving on all cells, and to detect subsets that might be missed otherwise. R code is available at https://github.com/SofieVG/FlowSOM and will be made available at Bioconductor.

ISAC's classification results file format

Identifying homogenous sets of cell populations in flow cytometry is an important process for sorting and selecting populations of interests for further data acquisition and analysis. Many computational methods are now available to automate this process, with several algorithms partitioning cells based on high-dimensional separation versus the traditional pairwise two-dimensional visualization approach of manual gating. ISAC's classification results file format was developed to exchange the results of both manual gating and algorithmic classification approaches in a standardized way based on per event based classifications, including the potential for soft classifications expressed as the probability of an event being a member of a class. © 2014 International Society for Advancement of Cytometry.

Lipid vesicle shape analysis from populations using light video microscopy and computer vision

We present a method for giant lipid vesicle shape analysis that combines manually guided large-scale video microscopy and computer vision algorithms to enable analyzing vesicle populations. The method retains the benefits of light microscopy and enables non-destructive analysis of vesicles from suspensions containing up to several thousands of lipid vesicles (1–50 µm in diameter). For each sample, image analysis was employed to extract data on vesicle quantity and size distributions of their projected diameters and isoperimetric quotients (measure of contour roundness). This process enables a comparison of samples from the same population over time, or the comparison of a treated population to a control. Although vesicles in suspensions are heterogeneous in sizes and shapes and have distinctively non-homogeneous distribution throughout the suspension, this method allows for the capture and analysis of repeatable vesicle samples that are representative of the population inspected.

flowDensity: reproducing manual gating of flow cytometry data by automated density-based cell population identification

SUMMARY

flowDensity facilitates reproducible, high-throughput analysis of flow cytometry data by automating a predefined manual gating approach. The algorithm is based on a sequential bivariate gating approach that generates a set of predefined cell populations. It chooses the best cut-off for individual markers using characteristics of the density distribution. The Supplementary Material is linked to the online version of the manuscript.

AVAILABILITY AND IMPLEMENTATION

R source code freely available through BioConductor (http://master.bioconductor.org/packages/devel/bioc/html/flowDensity.html.). Data available from FlowRepository.org (dataset FR-FCM-ZZBW).

CONTACT

rbrinkman@bccrc.ca

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

A crowd-sourcing approach for the construction of species-specific cell signaling networks

Motivation: Animal models are important tools in drug discovery and for understanding human biology in general. However, many drugs that initially show promising results in rodents fail in later stages of clinical trials. Understanding the commonalities and differences between human and rat cell signaling networks can lead to better experimental designs, improved allocation of resources and ultimately better drugs.

Results: The sbv IMPROVER Species-Specific Network Inference challenge was designed to use the power of the crowds to build two species-specific cell signaling networks given phosphoproteomics, transcriptomics and cytokine data generated from NHBE and NRBE cells exposed to various stimuli. A common literature-inspired reference network with 220 nodes and 501 edges was also provided as prior knowledge from which challenge participants could add or remove edges but not nodes. Such a large network inference challenge not based on synthetic simulations but on real data presented unique difficulties in scoring and interpreting the results. Because any prior knowledge about the networks was already provided to the participants for reference, novel ways for scoring and aggregating the results were developed. Two human and rat consensus networks were obtained by combining all the inferred networks. Further analysis showed that major signaling pathways were conserved between the two species with only isolated components diverging, as in the case of ribosomal S6 kinase RPS6KA1. Overall, the consensus between inferred edges was relatively high with the exception of the downstream targets of transcription factors, which seemed more difficult to predict.

Contact:ebilal@us.ibm.com or gustavo@us.ibm.com.

Supplementary information:Supplementary data are available at Bioinformatics online.

A non-parametric Bayesian model for joint cell clustering and cluster matching: identification of anomalous sample phenotypes with random effects

Background

Flow cytometry (FC)-based computer-aided diagnostics is an emerging technique utilizing modern multiparametric cytometry systems.

The major difficulty in using machine-learning approaches for classification of FC data arises from limited access to a wide variety of anomalous samples for training. In consequence, any learning with an abundance of normal cases and a limited set of specific anomalous cases is biased towards the types of anomalies represented in the training set. Such models do not accurately identify anomalies, whether previously known or unknown, that may exist in future samples tested. Although one-class classifiers trained using only normal cases would avoid such a bias, robust sample characterization is critical for a generalizable model. Owing to sample heterogeneity and instrumental variability, arbitrary characterization of samples usually introduces feature noise that may lead to poor predictive performance. Herein, we present a non-parametric Bayesian algorithm called ASPIRE (anomalous sample phenotype identification with random effects) that identifies phenotypic differences across a batch of samples in the presence of random effects. Our approach involves simultaneous clustering of cellular measurements in individual samples and matching of discovered clusters across all samples in order to recover global clusters using probabilistic sampling techniques in a systematic way.

Results

We demonstrate the performance of the proposed method in identifying anomalous samples in two different FC data sets, one of which represents a set of samples including acute myeloid leukemia (AML) cases, and the other a generic 5-parameter peripheral-blood immunophenotyping. Results are evaluated in terms of the area under the receiver operating characteristics curve (AUC). ASPIRE achieved AUCs of 0.99 and 1.0 on the AML and generic blood immunophenotyping data sets, respectively.

Conclusions

These results demonstrate that anomalous samples can be identified by ASPIRE with almost perfect accuracy without a priori access to samples of anomalous subtypes in the training set. The ASPIRE approach is unique in its ability to form generalizations regarding normal and anomalous states given only very weak assumptions regarding sample characteristics and origin. Thus, ASPIRE could become highly instrumental in providing unique insights about observed biological phenomena in the absence of full information about the investigated samples.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2105-15-314) contains supplementary material, which is available to authorized users.

Loss of blast heterogeneity in myelodysplastic syndrome and other chronic myeloid neoplasms

OBJECTIVES

Flow cytometry immunophenotyping has been suggested as an adjunctive technique in the evaluation of myeloid malignancies, especially in the myelodysplastic syndromes. However, its use has been limited due to complexity and cost restraints. The goal of this study is to attempt a simpler approach to flow cytometry immunophenotyping in myeloid neoplasms.

METHODS

We analyzed bone marrow specimens of 45 selected patients and an additional 99 consecutive random patients using a limited antibody panel.

RESULTS

Normal CD34-positive blasts show a characteristic pattern of CD13/HLA-DR expression, with three readily identifiable subpopulations. In contrast, myeloid neoplasms frequently show loss of this heterogeneity.

CONCLUSIONS

Analysis of a limited antibody panel with a focus on CD13/HLA-DR expression provides relatively high specificity and sensitivity for the detection of myeloid neoplasms.

Cell population identification using fluorescence-minus-one controls with a one-class classifying algorithm

MOTIVATION

The tried and true approach of flow cytometry data analysis is to manually gate on each biomarker separately, which is feasible for a small number of biomarkers, e.g. less than five. However, this rapidly becomes confusing as the number of biomarker increases. Furthermore, multivariate structure is not taken into account. Recently, automated gating algorithms have been implemented, all of which rely on unsupervised learning methodology. However, all unsupervised learning outputs suffer the same difficulties in validation in the absence of external knowledge, regardless of application domain.

RESULTS

We present a new semi-automated algorithm for population discovery that is based on comparison to fluorescence-minus-one controls, thus transferring the problem into that of one-class classification, as opposed to being an unsupervised learning problem. The novel one-class classification algorithm is based on common principal components and can accommodate complex mixtures of multivariate densities. Computational time is short, and the simple nature of the calculations means the algorithm can easily be adapted to process large numbers of cells (10(6)). Furthermore, we are able to find rare cell populations as well as populations with low biomarker concentration, both of which are inherently hard to do in an unsupervised learning context without prior knowledge of the samples' composition.

AVAILABILITY AND IMPLEMENTATION

R scripts are available via https://fccf.mpiib-berlin.mpg.de/daten/drfz/bioinformatics/with{username,password}={bioinformatics,Sar=Gac4}.

Intracellular staining for cytokines and transcription factors

Within the past years immune cells in general and T cells in particular have been categorized into a vast variety of subsets with different functional properties. One of the key technologies fueling this emerging complexity is intracellular staining for effector cytokines and/or lineage-defining transcription factors. Here we discuss the critical steps for performing successful multicolor immunophenotyping of mouse T cells in combination with analysis of intracellular molecules after ex vivo isolation.

Batch Discovery of Recurring Rare Classes toward Identifying Anomalous Samples

We present a clustering algorithm for discovering rare yet significant recurring classes across a batch of samples in the presence of random effects. We model each sample data by an infinite mixture of Dirichlet-process Gaussian-mixture models (DPMs) with each DPM representing the noisy realization of its corresponding class distribution in a given sample. We introduce dependencies across multiple samples by placing a global Dirichlet process prior over individual DPMs. This hierarchical prior introduces a sharing mechanism across samples and allows for identifying local realizations of classes across samples. We use collapsed Gibbs sampler for inference to recover local DPMs and identify their class associations. We demonstrate the utility of the proposed algorithm, processing a flow cytometry data set containing two extremely rare cell populations, and report results that significantly outperform competing techniques. The source code of the proposed algorithm is available on the web via the link: http://cs.iupui.edu/~dundar/aspire.htm.

CCAST: a model-based gating strategy to isolate homogeneous subpopulations in a heterogeneous population of single cells

A model-based gating strategy is developed for sorting cells and analyzing populations of single cells. The strategy, named CCAST, for Clustering, Classification and Sorting Tree, identifies a gating strategy for isolating homogeneous subpopulations from a heterogeneous population of single cells using a data-derived decision tree representation that can be applied to cell sorting. Because CCAST does not rely on expert knowledge, it removes human bias and variability when determining the gating strategy. It combines any clustering algorithm with silhouette measures to identify underlying homogeneous subpopulations, then applies recursive partitioning techniques to generate a decision tree that defines the gating strategy. CCAST produces an optimal strategy for cell sorting by automating the selection of gating markers, the corresponding gating thresholds and gating sequence; all of these parameters are typically manually defined. Even though CCAST is optimized for cell sorting, it can be applied for the identification and analysis of homogeneous subpopulations among heterogeneous single cell data. We apply CCAST on single cell data from both breast cancer cell lines and normal human bone marrow. On the SUM159 breast cancer cell line data, CCAST indicates at least five distinct cell states based on two surface markers (CD24 and EPCAM) and provides a gating sorting strategy that produces more homogeneous subpopulations than previously reported. When applied to normal bone marrow data, CCAST reveals an efficient strategy for gating T-cells without prior knowledge of the major T-cell subtypes and the markers that best define them. On the normal bone marrow data, CCAST also reveals two major mature B-cell subtypes, namely CD123+ and CD123- cells, which were not revealed by manual gating but show distinct intracellular signaling responses. More generally, the CCAST framework could be used on other biological and non-biological high dimensional data types that are mixtures of unknown homogeneous subpopulations.

NetFCM: a semi-automated web-based method for flow cytometry data analysis

Multi-parametric flow cytometry (FCM) represents an invaluable instrument to conduct single cell analysis and has significantly increased our understanding of the immune system. However, due to new techniques allowing us to measure an increased number of phenotypes within the immune system, FCM data analysis has become more complex and labor-intensive than previously. We have therefore developed a semi-automatic gating strategy (NetFCM) that uses clustering and principal component analysis (PCA) together with other statistical methods to mimic manual gating approaches. NetFCM is an online tool both for subset identification as well as for quantification of differences between samples. Additionally, NetFCM can classify and cluster samples based on multidimensional data. We tested the method using a data set of peripheral blood mononuclear cells collected from 23 HIV-infected individuals, which were stimulated with overlapping HIV Gag-p55 and CMV-pp65 peptides or medium alone (negative control). NetFCM clustered the virus-specific CD8+ T cells based on IFNγ and TNF responses into distinct compartments. Additionally, NetFCM was capable of identifying HIV- and CMV-specific responses corresponding to those obtained by manual gating strategies. These data demonstrate that NetFCM has the potential to identify relevant T cell populations by mimicking classical FCM data analysis and reduce the subjectivity and amount of time associated with such analysis. © 2014 International Society for Advancement of Cytometry.

Joint modeling and registration of cell populations in cohorts of high-dimensional flow cytometric data

In biomedical applications, an experimenter encounters different potential sources of variation in data such as individual samples, multiple experimental conditions, and multivariate responses of a panel of markers such as from a signaling network. In multiparametric cytometry, which is often used for analyzing patient samples, such issues are critical. While computational methods can identify cell populations in individual samples, without the ability to automatically match them across samples, it is difficult to compare and characterize the populations in typical experiments, such as those responding to various stimulations or distinctive of particular patients or time-points, especially when there are many samples. Joint Clustering and Matching (JCM) is a multi-level framework for simultaneous modeling and registration of populations across a cohort. JCM models every population with a robust multivariate probability distribution. Simultaneously, JCM fits a random-effects model to construct an overall batch template – used for registering populations across samples, and classifying new samples. By tackling systems-level variation, JCM supports practical biomedical applications involving large cohorts. Software for fitting the JCM models have been implemented in an R package EMMIX-JCM, available from http://www.maths.uq.edu.au/~gjm/mix_soft/EMMIX-JCM/.

Automated identification of stratifying signatures in cellular subpopulations

Elucidation and examination of cellular subpopulations that display condition-specific behavior can play a critical contributory role in understanding disease mechanism, as well as provide a focal point for development of diagnostic criteria linking such a mechanism to clinical prognosis. Despite recent advancements in single-cell measurement technologies, the identification of relevant cell subsets through manual efforts remains standard practice. As new technologies such as mass cytometry increase the parameterization of single-cell measurements, the scalability and subjectivity inherent in manual analyses slows both analysis and progress. We therefore developed Citrus (cluster identification, characterization, and regression), a data-driven approach for the identification of stratifying subpopulations in multidimensional cytometry datasets. The methodology of Citrus is demonstrated through the identification of known and unexpected pathway responses in a dataset of stimulated peripheral blood mononuclear cells measured by mass cytometry. Additionally, the performance of Citrus is compared with that of existing methods through the analysis of several publicly available datasets. As the complexity of flow cytometry datasets continues to increase, methods such as Citrus will be needed to aid investigators in the performance of unbiased--and potentially more thorough--correlation-based mining and inspection of cell subsets nested within high-dimensional datasets.

Multi-color flow cytometric immunophenotyping for detection of minimal residual disease in AML: past, present and future

Current chemotherapeutic regimens achieve CR in a large percentage of patients with AML. However, relapse after CR remains a significant problem. The presence of leukemic cells at levels too low to be detected by conventional microscopy, termed minimal residual disease (MRD), has been associated with an increased risk of relapse and shortened survival. Detection of MRD requires the use of highly sensitive ancillary techniques. Multi-color flow cytometric immunophenotyping is a sensitive method for quick and accurate detection of MRD. Use of this method in patient management may result in lower rates of relapse and improved survival, and is an effective means of assessing novel therapeutic agents. This method can be used in the vast majority of patients with AML, regardless of the immunophenotypic, cytogenetic and molecular genetic abnormalities present. Unfortunately, conflicting data regarding optimum methods of measurement and reporting, as well as the expertize required to interpret results have limited broad application of this technique. We provide a broad overview of this technique, including its advantages and limitations, and discuss the methods employed at our institution. We also review several possible areas of future investigation.

Setting objective thresholds for rare event detection in flow cytometry

The accurate identification of rare antigen-specific cytokine positive cells from peripheral blood mononuclear cells (PBMC) after antigenic stimulation in an intracellular staining (ICS) flow cytometry assay is challenging, as cytokine positive events may be fairly diffusely distributed and lack an obvious separation from the negative population. Traditionally, the approach by flow operators has been to manually set a positivity threshold to partition events into cytokine-positive and cytokine-negative. This approach suffers from subjectivity and inconsistency across different flow operators. The use of statistical clustering methods does not remove the need to find an objective threshold between between positive and negative events since consistent identification of rare event subsets is highly challenging for automated algorithms, especially when there is distributional overlap between the positive and negative events ("smear"). We present a new approach, based on the Fβ measure, that is similar to manual thresholding in providing a hard cutoff, but has the advantage of being determined objectively. The performance of this algorithm is compared with results obtained by expert visual gating. Several ICS data sets from the External Quality Assurance Program Oversight Laboratory (EQAPOL) proficiency program were used to make the comparisons. We first show that visually determined thresholds are difficult to reproduce and pose a problem when comparing results across operators or laboratories, as well as problems that occur with the use of commonly employed clustering algorithms. In contrast, a single parameterization for the Fβ method performs consistently across different centers, samples, and instruments because it optimizes the precision/recall tradeoff by using both negative and positive controls.

Unifying immunology with informatics and multiscale biology

The immune system is a highly complex and dynamic system. Historically, the most common scientific and clinical practice has been to evaluate its individual components. This kind of approach cannot always expose the interconnecting pathways that control immune-system responses and does not reveal how the immune system works across multiple biological systems and scales. High-throughput technologies can be used to measure thousands of parameters of the immune system at a genome-wide scale. These system-wide surveys yield massive amounts of quantitative data that provide a means to monitor and probe immune-system function. New integrative analyses can help synthesize and transform these data into valuable biological insight. Here we review some of the computational analysis tools for high-dimensional data and how they can be applied to immunology.

Understanding health and disease with multidimensional single-cell methods

Current efforts in the biomedical sciences and related interdisciplinary fields are focused on gaining a molecular understanding of health and disease, which is a problem of daunting complexity that spans many orders of magnitude in characteristic length scales, from small molecules that regulate cell function to cell ensembles that form tissues and organs working together as an organism. In order to uncover the molecular nature of the emergent properties of a cell, it is essential to measure multiple-cell components simultaneously in the same cell. In turn, cell heterogeneity requires multiple-cells to be measured in order to understand health and disease in the organism. This review summarizes current efforts towards a data-driven framework that leverages single-cell technologies to build robust signatures of healthy and diseased phenotypes. While some approaches focus on multicolor flow cytometry data and other methods are designed to analyze high-content image-based screens, we emphasize the so-called Supercell/SVM paradigm (recently developed by the authors of this review and collaborators) as a unified framework that captures mesoscopic-scale emergence to build reliable phenotypes. Beyond their specific contributions to basic and translational biomedical research, these efforts illustrate, from a larger perspective, the powerful synergy that might be achieved from bringing together methods and ideas from statistical physics, data mining, and mathematics to solve the most pressing problems currently facing the life sciences.

SWIFT-scalable clustering for automated identification of rare cell populations in large, high-dimensional flow cytometry datasets, part 2: biological evaluation

A multistage clustering and data processing method, SWIFT (detailed in a companion manuscript), has been developed to detect rare subpopulations in large, high-dimensional flow cytometry datasets. An iterative sampling procedure initially fits the data to multidimensional Gaussian distributions, then splitting and merging stages use a criterion of unimodality to optimize the detection of rare subpopulations, to converge on a consistent cluster number, and to describe non-Gaussian distributions. Probabilistic assignment of cells to clusters, visualization, and manipulation of clusters by their cluster medians, facilitate application of expert knowledge using standard flow cytometry programs. The dual problems of rigorously comparing similar complex samples, and enumerating absent or very rare cell subpopulations in negative controls, were solved by assigning cells in multiple samples to a cluster template derived from a single or combined sample. Comparison of antigen-stimulated and control human peripheral blood cell samples demonstrated that SWIFT could identify biologically significant subpopulations, such as rare cytokine-producing influenza-specific T cells. A sensitivity of better than one part per million was attained in very large samples. Results were highly consistent on biological replicates, yet the analysis was sensitive enough to show that multiple samples from the same subject were more similar than samples from different subjects. A companion manuscript (Part 1) details the algorithmic development of SWIFT. © 2014 The Authors. Published by Wiley Periodicals Inc.

Single-cell technologies for monitoring immune systems

The complex heterogeneity of cells, and their interconnectedness with each other, are major challenges to identifying clinically relevant measurements that reflect the state and capability of the immune system. Highly multiplexed, single-cell technologies may be critical for identifying correlates of disease or immunological interventions as well as for elucidating the underlying mechanisms of immunity. Here we review limitations of bulk measurements and explore advances in single-cell technologies that overcome these problems by expanding the depth and breadth of functional and phenotypic analysis in space and time. The geometric increases in complexity of data make formidable hurdles for exploring, analyzing and presenting results. We summarize recent approaches to making such computations tractable and discuss challenges for integrating heterogeneous data obtained using these single-cell technologies.

Beyond model antigens: high-dimensional methods for the analysis of antigen-specific T cells

Adaptive immune responses often begin with the formation of a molecular complex between a T-cell receptor (TCR) and a peptide antigen bound to a major histocompatibility complex (MHC) molecule. These complexes are highly variable, however, due to the polymorphism of MHC genes, the random, inexact recombination of TCR gene segments, and the vast array of possible self and pathogen peptide antigens. As a result, it has been very difficult to comprehensively study the TCR repertoire or identify and track more than a few antigen-specific T cells in mice or humans. For mouse studies, this had led to a reliance on model antigens and TCR transgenes. The study of limited human clinical samples, in contrast, requires techniques that can simultaneously survey TCR phenotype and function, and TCR reactivity to many T-cell epitopes. Thanks to recent advances in single-cell and cytometry methodologies, as well as high-throughput sequencing of the TCR repertoire, we now have or will soon have the tools needed to comprehensively analyze T-cell responses in health and disease.