Cytometry Part A

IntroductionThe advent of full spectral flow cytometry has enabled the development of complex panels with over 35 colors, with the latest panels reaching 50 colors (1). This capability is made possible by cytometers equipped with numerous detectors beyond those in traditional cytometers and an expanded range of fluorochromes with emission peaks across the visible spectrum. However, our observations reveal significant challenges in the current unmixing, spread prediction, and panel design methodologies. Existing tools and guidelines, largely optimized for panels with up to 20+ colors, are limited in their ability to navigate this new ultra-high-color landscape. Without improvements in unmixing algorithms, predictive tools for spread, and design strategies, researchers risk creating suboptimal panels and obtaining inaccurate results. This article aims to highlight a range of emerging challenges associated with ultra-high parameter flow cytometry, particularly for practitioners accustomed to conventional panel design and analysis. As the field advances toward increasingly complex multiparameter experiments, novel issues have surfaced--many of which were previously unrecognized. Although this work does not provide comprehensive solutions to all of these observations, it underscores the need for continued methodological development. We anticipate that ongoing research by experts in the field will yield robust frameworks to address these challenges and advance best practices in high-dimensional cytometric analysis. Brief descriptionO_LIDifferent unmixing/compensation algorithms can result in different biological interpretations from the same raw dataset. C_LIO_LIA method to identify the optimal unmixing algorithm for accurate analysis is discussed. C_LIO_LIBoth panel size and specific fluorochrome combinations significantly impact population spread. C_LIO_LISSM (Spillover Spreading Matrix) values are influenced by panel size and fluorochrome combinations, which, if not carefully evaluated, may lead to misleading conclusions during panel design. C_LI

We report the development of an optimized 50-color spectral flow cytometry panel designed for the in-depth analysis of the immune system in human blood and tissues, with the goal of maximizing the amount of information that can be collected using currently available flow cytometry platforms. We established and tested this panel using peripheral blood mononuclear cells (PBMCs), but included CD45 to enable its use for the analysis of human tissue samples. The panel contains lineage markers for all major immune cell subsets, and an extensive set of phenotyping markers focused on the activation and differentiation status of the T cell and dendritic cell (DC) compartment. We outline the biological insight that can be gained from the simultaneous measurement of such a large number of proteins and propose that this approach provides a unique opportunity for the comprehensive exploration of the immune status in tissue biopsies and other human samples with a limited number of cells. Of note, we tested the panel to be compatible with cell sorting for further downstream applications. Furthermore, to facilitate the wide-spread implementation of such a panel across different cohorts and samples, we established a trimmed-down 45-color version which can be used with different spectral cytometry platforms. Finally, to generate this panel, we utilized not only existing panel design guidelines, but also developed new metrics to systematically identify the optimal combination of 50 fluorochromes and evaluate fluorochrome-specific resolution in the context of a 50-color unmixing matrix.

Flow cytometry analysis is widespread practice in cell biology, immunology and hematology. Cell populations of interest are typically identified by consecutively examining the expression levels of protein marker pairs. Since this manual gating process lacks standardization and is time-consuming, several machine learning (ML) methods for the automated gating of flow cytometry data have been proposed in recent years. We evaluated state-of-the-art ML methods for automated gating based on three criteria: gating performance in comparison to manual expert annotations, interpretability of the output, and feasibility of deployment in a clinical setting. Based on these criteria, we selected the top-performing methods and made them easily accessible in the Python package FLAG-X ("flow cytometry automated gating toolbox"), which further features a novel hybrid workflow that interlocks manual and automated gating and integrates seamlessly with standard software for manual gating procedure. To demonstrate its practical utility, we applied FLAG-X to representative cases from routine clinical practice. FLAG-X is available at https://anaconda.org/channels/bioconda/packages/flagx/overview.

The need for more in-depth exploration of the human immune system has moved the flow cytometry field forward with advances in instrumentation, reagent development and user-friendly implementations of data analysis methods. The increase in the number of markers evaluated simultaneously requires a careful selection of highly overlapping dyes to avoid introducing detrimental spread and compromising population resolution. In this manuscript, we present the strategy used in the development of a high-quality human 45-color panel which allows for comprehensive characterization of major cell lineages present in circulation including T cells, gamma delta T cells, NKT-like cells, B cells, NK cells, monocytes, basophils, dendritic cells, and ILCs, as well as more in-depth characterization of memory T cells. The steps taken to ensure that each marker in the panel was optimally resolved are discussed in detail. We highlight the outstanding discernment of cell activation, exhaustion, memory, and differentiation states of CD4+ and CD8+ T cells using this 45-color panel, enabling an in-depth description of very distinct phenotypes associated with the complexity of the T cell memory response. Furthermore, we present how this panel can be effectively used for cell sorting on instruments with a similar optical layout to achieve the same level of resolution. Functional evaluation of sorted specific rare cell subsets demonstrated significantly different patterns of immunological responses to stimulation, supporting functional and phenotypic differences within the T cell memory subsets. In summary, the combination of flow cytometry full spectrum technology, careful assay design and optimization, results in high resolution multiparametric assays. This approach offers the opportunity to fully characterize immunological profiles present in peripheral blood in the context of infectious diseases, autoimmunity, neurodegeneration, immunotherapy, and biomarker discovery. PURPOSE AND APPROPRIATE SAMPLE TYPESThis 45-color flow cytometry-based panel was developed as an expansion of the previously published OMIP-069 [1] and serves as an in-depth immunophenotyping of the major cell subsets present in human peripheral blood. The goal of this panel is to maximize the amount of high-quality data that can be acquired from a single sample, not only for more in-depth characterization of the immune system, but also to address the issue of limited sample availability. The panels development included identifying fluorochromes that could improve the performance of the original 40-color panel and expanding the number of markers for deeper delineation of memory status of T cell subpopulations. To increase the number of markers, it was critical that any expansion did not negatively impact the resolution and quality of the data. To achieve this, the fluorochrome combinations were carefully characterized to ensure optimal resolution of each marker. The panel allows for deep characterization of the major cell lineages present in circulation (CD4 T cells, CDS T cells, regulatory T cells, yo T cells, NKT-like cells, B cells, NK (Natural Killer) cells, monocytes, and dendritic cells), while also providing an in-depth characterization of the T cell compartment, with a combination of activation, inhibitory, exhaustion, and differentiation markers. The panel supports deep exploration of the memory status of CD4+ T cells, CDS+ T cells, and NKT-like cells. The steps taken in the optimization of the panel ensured outstanding resolution of each marker within the multicolor panel and unequivocal identification of each cell subset. This panel design and optimization will enhance the ability to characterize immunological profiles present in peripheral blood in the context of oncology, infectious diseases, autoimmunity, neurodegeneration, immunotherapy, and biomarker discovery. The panel was developed using fresh and cryopreserved human peripheral blood mononuclear cells (PBMCs) from healthy adults. We have not tested the panel on whole blood or biopsies; hence it is anticipated that the panel might require further optimization to be used with other sample types.

Sample multiplexing in flow cytometry is a powerful technique which allows for reduction of error, inclusion of control samples for batch effect correction, and reduction in both time and consumable usage. Current industry standard for barcoding in mass cytometry is an intracellular reagent, which requires fixation and permeabilization of sample prior to barcoding. We developed a barcode using the ubiquitous and well-tolerated membrane labeling lectin, wheat germ agglutinin. This barcode effectively labels all tested cell types, both live and fixed. We determine that barcode yields, or the ratio of debarcoded cells to total input cells, is stable in live pooled sample for at least an hour. This barcode does not show differential performance across major PBMC lineages. Thus, this universal wheat germ agglutinin-based barcode represents an advance in gentle, non-reactive cell surface barcoding for live cells.

Distinguishing cell types in peripheral blood smears is critical for diagnosing blood diseases, such as leukemia subtypes. Artificial intelligence can assist in automating cell classification. For training robust machine learning algorithms, however, large and well-annotated single-cell datasets are pivotal. Here, we introduce a large, publicly available, annotated peripheral blood dataset comprising >40,000 single-cell images classified into 18 classes by cytomorphology experts from the Munich Leukemia Laboratory, the largest European laboratory for blood disease diagnostics. By making our dataset publicly available, we provide a valuable resource for medical and machine learning researchers and support the development of reliable and clinically relevant diagnostic tools for diagnosing hematological diseases.

Although spectral flow cytometry has become a ubiquitous tool for cell analysis, the use of spectral cytometry on cell sorters requires additional considerations arising from the unique requirements of sorting workflows. Here, we show that care should be taken when ascertaining the purity of a sort on a spectral cell sorter, as the mismatch of buffers used for initial sample suspension and the buffers used for sort collection can affect the unmixing of the data, potentially giving rise to erroneous purity check results.

Full spectrum flow cytometry is a powerful tool for immune monitoring on a single-cell level and with currently available machines, panels of 40 or more markers per sample are possible. However, with an increased panel size, spectral unmixing issues arise, and appropriate single stain reference controls are required for accurate experimental results and to avoid unmixing errors. In contrast to conventional flow cytometry, full spectrum flow cytometry takes into account even minor differences in spectral signatures and requires the full spectrum of each fluorochrome to be identical in the reference control and the fully stained sample to ensure accurate and reliable results. In general, using the cells of interest is considered optimal, but certain markers may not be expressed at sufficient levels to generate a reliable positive control. In this case, compensation beads show some significant advantages as they bind a consistent amount of antibody independent of its specificity. In this study, we evaluated two types of manufactured compensation beads for use as reference controls for full spectrum cytometry and compared them to human and murine primary leukocytes. While most fluorochromes show the same spectral profile on beads and cells, we demonstrate that specific fluorochromes show a significantly different spectral profile depending on which type of compensation beads is used, and some fluorochromes should be used on cells exclusively. Finally, we provide a list of appropriate reference controls for 30 of the most commonly used and commercially available fluorochromes.

Flow cytometry is a powerful technology for high-throughput protein quantification at the single-cell level, widely used in basic research and routine clinical diagnostics. Traditionally, data analysis is carried out using manual gating, in which cut-offs are defined manually for each marker. Recent technical advances, including the introduction of mass cytometry, have increased the number of proteins that can be simultaneously assessed in each cell. To tackle the resulting escalation in data complexity, numerous new analysis algorithms have been developed. However, many of these show limitations in terms of providing statistical testing, data sharing, cross-experiment comparability integration with clinical data. We developed MetaGate as a platform for interactive statistical analysis and visualization of manually gated high-dimensional cytometry data with integration of clinical meta data. MetaGate allows manual gating to take place in traditional cytometry analysis software, while providing a combinatorial gating system for simple and transparent definition of biologically relevant cell populations. We demonstrate the utility of MetaGate through a comprehensive analysis of peripheral blood immune cells from 28 patients with diffuse large B-cell lymphoma (DLBCL) along with 17 age- and sex-matched healthy controls using two mass cytometry panels made of a total of 55 phenotypic markers. In a two-step process, raw data from 143 FCS files is first condensed through a data reduction algorithm and combined with information from manual gates, user-defined cellular populations and clinical meta data. This results in one single small project file containing all relevant information to allow rapid statistical calculation and visualization of any desired comparison, including box plots, heatmaps and volcano plots. Our detailed characterization of the peripheral blood immune cell repertoire in patients with DLBCL corroborate previous reports showing expansion of monocytic myeloid-derived suppressor cells, as well as an inverse correlation between NK cell numbers and disease progression.

Identifying cell populations in flow cytometry data is mostly done via a "manual gating" method that often lacks verifiability and reproducibility, even in the hands of experienced investigators. Recently developed automatic gating methods have been shown to have good performance in cell population identification, but may require fine-tuned setup from experts or struggle to identify small populations. Here, we introduce an easily trainable multilayer perceptron neural network for automatic gating (MLPgater). Compared to the three popular automatic gating methods LDA, FlowSOM and PhenoGraph on three mass and six fluorescence cytometry datasets, MLPgater is most accurate by a substantial margin, tied with LDA as the fastest and uniquely able to replace manual gating except for training purposes. Furthermore, we show that combining MLPgater with UMAPs guided dimensionality reduction feature and DBMs clustering (MUDflow) effectively detects new populations that did not exist or were not identified in the training set.

To achieve accurate and reproducible cytometry data analysis, we benchmarked 19 machine learning algorithms for supervised and unsupervised cell classification. The underlying data encompassed 138 million cells from seven independent datasets including conventional flow cytometry, spectral flow cytometry and mass cytometry. We found that tree-based classifiers and in particular Decision Trees, outperformed other approaches in classification accuracy, speed and memory use. High accuracy was achieved even for cell populations rarer than 1% using decision trees. We validated our decision tree-based approach in a clinical setting using diagnostic blood T cell phenotyping of 107 patients. Automatic quantification of CD4 helper T cell phenotypes achieved 99 % accuracy compared to manual expert assessment. Finally, we combined automated data transformation, supervised and unsupervised gating, an application program interface and a user-friendly desktop-application into FACSPy and FACSPyUI, a fast and scalable open-source toolbox for the analysis and visualization of cytometry data.

In the past decade, high-dimensional single cell technologies have revolutionized basic and translational immunology research and are now a key element of the toolbox used by scientists to study the immune system. However, analysis of the data generated by these approaches often requires clustering algorithms and dimensionality reduction representation which are computationally intense and difficult to evaluate and optimize. Here we present Cyclone, an analysis pipeline integrating dimensionality reduction, clustering, evaluation and optimization of clustering resolution, and downstream visualization tools facilitating the analysis of a wide range of cytometry data. We benchmarked and validated Cyclone on mass cytometry (CyTOF), full spectrum fluorescence-based cytometry, and multiplexed immunofluorescence (IF) in a variety of biological contexts, including infectious diseases and cancer. In each instance, Cyclone not only recapitulates gold standard immune cell identification, but also enables the unsupervised identification of lymphocytes and mononuclear phagocytes subsets that are associated with distinct biological features. Altogether, the Cyclone pipeline is a versatile and accessible pipeline for performing, optimizing, and evaluating clustering on variety of cytometry datasets which will further power immunology research and provide a scaffold for biological discovery.

Introduction The complete blood count with differential (CBD) is one of the most commonly performed blood tests worldwide, used in nearly all areas of medicine. Although modern CBD analyzers generate flow cytometry based single cell measurements,the resultant CBD markers are limited to coarse summary features, such as total cell counts and average cell sizes. This means, the markers cannotdetect subtle cell population shifts that may signal early stage pathogenesis. To test this, we evaluate whether AI based analysis of the raw single cell data underlying the CBD can be used to develop novel, clinically prognostic biomarkers, across patient settings. Method We developed two complementary methods for biomarker discovery using CBD tests and evaluated them with longitudinal data from an academic medical center. To create interpretable biomarkers, we clustered cells into physiologically meaningful subpopulations and performed robust statistical summarization. In tandem, self supervised autoencoders were developed to extract novel nonlinear markers. We evaluated the utility of these clustering (CLS) and autoencoder (AE) markers for patient prognostication across a range of outcomes (mortality, inpatient admission, and future disease development). Results Our study included 242,623 CBD samples from 127,545 patients. Both clustering and embedding approaches successfully generated hundreds of new clinical biomarkers. Many biomarkers showed strong prognostic associations for all cause mortality, inpatient admission, and development of anemia, cancer, or cardiovascular disease, with associations remaining significant after adjustment for demographics and clinical CBD markers. A large subset of these prognostic markers also showed high novelty, having low correlations to existing CBD markers, while also exhibiting significant correlations with broader physiologic signals, such as inflammatory, hormonal, infectious, and coagulopathic markers. Conclusion Collectively, these results demonstrate how modern AI techniques can allow for deeper phenotyping of routine clinical blood counts, generating novel biomarkers that capture more subtle physiologic signals than what are currently clinically utilized.

Advances in spectral cytometry instrumentation and fluorescent reagents have led to the possibility of ultra-high-parameter panels exceeding 50 colors. However, panel size is limited in practice by unmixing-dependent spreading (UDS), a mathematical phenomenon which leads to a progressive deterioration of unmixed signal-to-noise ratios in panels that contain fluorochrome combinations with significant spectral overlap. Choosing spectrally compatible sets of fluorochromes that avoid UDS is a complex and labor-intensive task involving substantial trial-and-error experimentation. Here, we provide a detailed explanation of UDS and practical strategies for handling UDS in large spectral panels. We describe the empirical hallmarks of UDS, demonstrate how to quantify its impact, and dissect its underlying mathematical cause in terms of spectral collinearity. We present novel computational metrics that can be used to select optimal combinations of fluorochromes in a platform-agnostic fashion based on publicly available reference data, providing a general tool for spectral panel design.

AI-based image recognition has significantly advanced the analysis of tissues and individual cells both in the context of translational studies and diagnostics. To date, recognition is primarily based on the identification of certain cell characteristics (e.g. by staining). The morphological assessment of unstained cells holds additional potential, as it allows for virtually real-time assessment without the need to manipulate the cells. This facilitates longitudinal observations, as required for drug testing, and forms a basis for autonomous experimental execution. A semi-automated cell culture system (AICE3, LabMaite) was used to culture myeloid leukemic cell lines (K562, HL-60, Kasumi-1). K562 cells were treated with hemin and PMA to induce erythroid and megakaryocytic differentiation, respectively. Cell images were acquired using automated bright field microscopy. Images were used to train an AI model using an NVIDIA DGX A100 GPU with Ultralytics YOLOv8. Morphologic features were extracted using RedTell. The model reliably distinguished K562 cells from HL-60 and Kasumi-1 using >400 images per class (average >15 cells/image). Bounding boxes were generated correctly (mAP@.5 >98%); precision and sensitivity exceeded 97%. Validation on an external K562 dataset confirmed these results. Classification of all three cell lines achieved >97% sensitivity/specificity and 94.6% precision. To test drug response, we used YOLOv8-s to distinguish untreated K562 cells from those undergoing erythroid or megakaryocytic differentiation (n >3,000 annotations). Precision, sensitivity, and specificity were >95%. RedTell identified 3 of 74 morphological traits contributing significantly to class separation. We demonstrate accurate, near real-time detection of unstained cells, enabling future AI-based drug testing.

In conventional flowcytometry one detector (primary) is dedicated for one fluorochrome. However, photons usually end up in other detectors too (fluorescence spillover). Compensation is a process that corrects the spillover signal from all detectors except the primary detector. Post compensation, the photon counting error of spillover signals become evident as spreading of the data. The spreading induced by spillover impairs the ability to resolve stained cell population from the unstained one, potentially reducing or completely losing cell populations. For successful multi-color panel design, it is important to know the expected spillover to maximize the data resolution. The Spillover Spreading Matrix (SSM) can be used to estimate the spread, but the outcome is dependent on detector sensitivity. Simply, the same single stained sample produces different spillover spread values when detector(s) sensitivity is altered. Many researchers mistakenly use this artifact to "reduce" the spread by decreasing detector sensitivity. This can result in diminished capacity to resolve dimly expressing cell populations. Here, we introduce SQI (Spread Quantification Index), that can quantify the spillover spread independent of detector sensitivity and independent of dynamic range. This allows users to compare spillover spread between instruments having different types of detectors, which is not possible using SSM.

Eosinophils are a rare immune cell subset with important roles in Th2 immunity and, recently, in cancer. Interleukin IL-33 (IL-33) is well recognized for its important roles in the activation of eosinophils in Th2 immunity. On the other hand, IL-33 has been recently discovered to play central roles in cancer, in particular by activating eosinophils and increase their degranulation consequent to an intrinsic tumor cell killing function. We propose a dual approach methodology to extrapolate functional interactions of eosinophils with tumor cells, as a result of eosinophil stimulation. Human eosinophils (Eos) isolated from the blood of healthy donors by dextran sedimentation followed by magnetic sorting are exposed to IL-33 (Eos33) or IL-5 (Eos5, control) for 18 h. These pre-conditioned cells are then co-cultured with A375P melanoma cells to monitor cell-cell interactions. Acoustic focusing flow cytometry analysis is employed to evaluate the presence of Eos-tumor cell conjugates after 1h incubation of human eosinophils and A375P melanoma cells. Moreover, a 24 h time-lapse video recording approach is employed to obtain single cell tracking Eos profiles. This allows to quantitatively determine the interaction extent of Eos33, as opposed to Eos5 (control condition), with tumor cells. In conclusion, our protocols easily and quickly allow the extrapolation of relevant kinematic and biologically relevant parameters for tumor reactive eosinophils. Furthermore, these methods are adaptable to various models with other types of immune cell subsets and cancer cells and can be implemented on different video microscopy platforms and advanced flow cytometry systems.

Over the past decade, there has been an explosion in the characterisation and discovery of cell populations using single-cell technologies. Single-cell multi-omics data, particularly those incorporating gene and protein expression, are increasingly commonplace and can lead to more refined characterisation of cell types. A common challenge for biologists is to isolate cells of interest using a minimal number of markers for cytometry experiments. Although several methods exist for marker selection, there is limited guidance on the relative performance of these methods, and a wrapper package that combines multiple methods is lacking. The method that performs best can vary depending on the dataset and it can be challenging for researchers to test multiple methods for a given dataset. To address these issues, we present MiniMarS (Minimal Marker Selection), an R package that serves as a wrapper for 10 different algorithms. It allows users to determine the best-performing algorithm for identifying the optimal number of markers that will delineate cell populations in their dataset. MiniMarS uses pre-annotated cells with protein features from CyTOF or sequencing-based assays such as CITE-seq and Abseq as input. Outputs include 1) the minimum number of protein markers required to identify the annotated cell populations using a range of marker selection algorithms, and 2) a range of metrics to evaluate the performance of each algorithm. MiniMarS effectively differentiated populations across various datasets, including those from human blood, bone marrow, thymus, mouse spleen, and lymph nodes, even after subsampling over 41,000 cells to 2,500 cells. MiniMarS also identified 15 markers from CITE-seq data, which were then used to successfully identify the same 11 cell subsets in a CyTOF dataset (F1 score>0.9). Additionally, we showed that by appropriately combining clusters, MiniMarS improves the F1 score of a rare population identification (<1% of total cells) by 28.7%. Together, these findings highlight the broad applicability of MiniMarS in identifying appropriate markers for distinguishing cell populations.

Technical artifacts that occur during the data acquisition process of cytometry data can result in erroneous data. We showed the presence of these data leads to biased gating analysis. Common technical issues, such as clogging, can cause spurious events and fluorescence intensity shifting. These events should be identified and potentially removed before being passed to the next stage of the gating analysis. flowCut, an R package, automatically detects anomaly events and flags files for flow cytometry experiments. flowCut outperforms existing automated approaches in our evaluation. flowCut is available as an R package at: https://github.com/jmeskas/flowCut. Test data uploaded to FlowRepository (Repository ID: FR-FCM-ZYPD) along with the manual results.

BackgroundErythrocyte indices are essential for the diagnosis and monitoring of hematologic diseases, but their determination depends on automated hematology analyzers, which limits access in regions with limited laboratory infrastructure. Although artificial intelligence approaches have been proposed for hematologic analysis, they usually rely on slide scanners or digitization systems. To date, no validated approaches have been identified in the literature that estimate these indices directly from images obtained through the eyepiece of conventional optical microscopes. ObjectiveTo evaluate the feasibility of automated prediction of erythrocyte indices from blood smear images obtained directly through the eyepiece of conventional microscopes using convolutional neural networks. MethodsTwo hundred blood samples stained using the May-Grunwald-Giemsa method were analyzed and photographed using a standard optical microscope. Four architectures, DenseNet-121, EfficientNet-B0, ResNet-18, and ResNet-34, were evaluated at different resolutions using 10-fold K-Fold cross-validation. ResultsFor RBC, HGB, and HCT, ResNet-34 at a resolution of 1024x1024 pixels achieved superior performance, with R2 between 0.90 and 0.92, Pearson correlation r > 0.95, and mean absolute errors of 0.184 x106/{micro}L, 0.524 g/dL and 1.292%, respectively. For RDW-CV, DenseNet-121 achieved R2 = 0.49 and r = 0.71, reflecting the greater complexity of this parameter. Bland-Altman analysis confirmed adequate agreement, with biases close to zero and more than 94% of observations within the limits of agreement. ConclusionArtificial intelligence demonstrated excellent predictive performance in estimating the erythrocyte indices RBC, HGB, and HCT, with R2 > 0.90, from images obtained using a conventional microscope and accessible hardware. This approach has significant potential to democratize access to hematologic analysis in resource-limited settings, although multicenter validation and regulatory evaluation are required before clinical implementation.