Efficient cytometry analysis with FlowSOM in Python boosts interoperability with other single-cell tools.

MOTIVATION: We describe a new Python implementation of FlowSOM, a clustering method for cytometry data. RESULTS: This implementation is faster than the original version in R, better adapted to work with single-cell omics data including integration with current single-cell data structures and includes all the original visualizations, such as the star and pie plot. AVAILABILITY AND IMPLEMENTATION: The FlowSOM Python implementation is freely available on GitHub: https://github.com/saeyslab/FlowSOM_Python.

IRF8 Transcription Factor Controls Survival and Function of Terminally Differentiated Conventional and Plasmacytoid Dendritic Cells, Respectively.

Interferon regulatory factor-8 (IRF8) has been proposed to be essential for development of monocytes, plasmacytoid dendritic cells (pDCs) and type 1 conventional dendritic cells (cDC1s) and remains highly expressed in differentiated DCs. Transcription factors that are required to maintain the identity of terminally differentiated cells are designated "terminal selectors." Using BM chimeras, conditional Irf8(fl/fl) mice and various promotors to target Cre recombinase to different stages of monocyte and DC development, we have identified IRF8 as a terminal selector of the cDC1 lineage controlling survival. In monocytes, IRF8 was necessary during early but not late development. Complete or late deletion of IRF8 had no effect on pDC development or survival but altered their phenotype and gene-expression profile leading to increased T cell stimulatory function but decreased type 1 interferon production. Thus, IRF8 differentially controls the survival and function of terminally differentiated monocytes, cDC1s, and pDCs.

Unsupervised High-Dimensional Analysis Aligns Dendritic Cells across Tissues and Species.

Dendritic cells (DCs) are professional antigen-presenting cells that hold great therapeutic potential. Multiple DC subsets have been described, and it remains challenging to align them across tissues and species to analyze their function in the absence of macrophage contamination. Here, we provide and validate a universal toolbox for the automated identification of DCs through unsupervised analysis of conventional flow cytometry and mass cytometry data obtained from multiple mouse, macaque, and human tissues. The use of a minimal set of lineage-imprinted markers was sufficient to subdivide DCs into conventional type 1 (cDC1s), conventional type 2 (cDC2s), and plasmacytoid DCs (pDCs) across tissues and species. This way, a large number of additional markers can still be used to further characterize the heterogeneity of DCs across tissues and during inflammation. This framework represents the way forward to a universal, high-throughput, and standardized analysis of DC populations from mutant mice and human patients.

Single-cell molecular profiling using ex vivo functional readouts fuels precision oncology in glioblastoma.

BACKGROUND: Functional profiling of freshly isolated glioblastoma (GBM) cells is being evaluated as a next-generation method for precision oncology. While promising, its success largely depends on the method to evaluate treatment activity which requires sufficient resolution and specificity. METHODS: Here, we describe the 'precision oncology by single-cell profiling using ex vivo readouts of functionality' (PROSPERO) assay to evaluate the intrinsic susceptibility of high-grade brain tumor cells to respond to therapy. Different from other assays, PROSPERO extends beyond life/death screening by rapidly evaluating acute molecular drug responses at single-cell resolution. RESULTS: The PROSPERO assay was developed by correlating short-term single-cell molecular signatures using mass cytometry by time-of-flight (CyTOF) to long-term cytotoxicity readouts in representative patient-derived glioblastoma cell cultures (n = 14) that were exposed to radiotherapy and the small-molecule p53/MDM2 inhibitor AMG232. The predictive model was subsequently projected to evaluate drug activity in freshly resected GBM samples from patients (n = 34). Here, PROSPERO revealed an overall limited capacity of tumor cells to respond to therapy, as reflected by the inability to induce key molecular markers upon ex vivo treatment exposure, while retaining proliferative capacity, insights that were validated in patient-derived xenograft (PDX) models. This approach also allowed the investigation of cellular plasticity, which in PDCLs highlighted therapy-induced proneural-to-mesenchymal (PMT) transitions, while in patients' samples this was more heterogeneous. CONCLUSION: PROSPERO provides a precise way to evaluate therapy efficacy by measuring molecular drug responses using specific biomarker changes in freshly resected brain tumor samples, in addition to providing key functional insights in cellular behavior, which may ultimately complement standard, clinical biomarker evaluations.

TAO-kinase 3 governs the terminal differentiation of NOTCH2-dependent splenic conventional dendritic cells.

Antigen-presenting conventional dendritic cells (cDCs) are broadly divided into type 1 and type 2 subsets that further adapt their phenotype and function to perform specialized tasks in the immune system. The precise signals controlling tissue-specific adaptation and differentiation of cDCs are currently poorly understood. We found that mice deficient in the Ste20 kinase Thousand and One Kinase 3 (TAOK3) lacked terminally differentiated ESAM+ CD4+ cDC2s in the spleen and failed to prime CD4+ T cells in response to allogeneic red-blood-cell transfusion. These NOTCH2- and ADAM10-dependent cDC2s were absent selectively in the spleen, but not in the intestine of Taok3-/- and CD11c-cre Taok3fl/fl mice. The loss of splenic ESAM+ cDC2s was cell-intrinsic and could be rescued by conditional overexpression of the constitutively active NOTCH intracellular domain in CD11c-expressing cells. Therefore, TAOK3 controls the terminal differentiation of NOTCH2-dependent splenic cDC2s.

Challenges in translational machine learning.

Machine learning (ML) algorithms are increasingly being used to help implement clinical decision support systems. In this new field, we define as "translational machine learning", joint efforts and strong communication between data scientists and clinicians help to span the gap between ML and its adoption in the clinic. These collaborations also improve interpretability and trust in translational ML methods and ultimately aim to result in generalizable and reproducible models. To help clinicians and bioinformaticians refine their translational ML pipelines, we review the steps from model building to the use of ML in the clinic. We discuss experimental setup, computational analysis, interpretability and reproducibility, and emphasize the challenges involved. We highly advise collaboration and data sharing between consortia and institutes to build multi-centric cohorts that facilitate ML methodologies that generalize across centers. In the end, we hope that this review provides a way to streamline translational ML and helps to tackle the challenges that come with it.

PeacoQC: Peak-based selection of high quality cytometry data.

In cytometry analysis, a large number of markers is measured for thousands or millions of cells, resulting in high-dimensional datasets. During the measurement of these samples, erroneous events can occur such as clogs, speed changes, slow uptake of the sample etc., which can influence the downstream analysis and can even lead to false discoveries. As these issues can be difficult to detect manually, an automated approach is recommended. In order to filter these erroneous events out, we created a novel quality control algorithm, Peak Extraction And Cleaning Oriented Quality Control (PeacoQC), that allows for automated cleaning of cytometry data. The algorithm will determine density peaks per channel on which it will remove low quality events based on their position in the isolation tree and on their mean absolute deviation distance to these density peaks. To evaluate PeacoQC's cleaning capability, it was compared to three other existing quality control algorithms (flowAI, flowClean and flowCut) on a wide variety of datasets. In comparison to the other algorithms, PeacoQC was able to filter out all different types of anomalies in flow, mass and spectral cytometry data, while the other methods struggled with at least one type. In the quantitative comparison, PeacoQC obtained the highest median balanced accuracy and a similar running time compared to the other algorithms while having a better scalability for large files. To ensure that the parameters chosen in the PeacoQC algorithm are robust, the cleaning tool was run on 16 public datasets. After inspection, only one sample was found where the parameters should be further optimized. The other 15 datasets were analyzed correctly indicating a robust parameter choice. Overall, we present a fast and accurate quality control algorithm that outperforms existing tools and ensures high-quality data that can be used for further downstream analysis. An R implementation is available.

Presence of innate lymphoid cells in allogeneic hematopoietic grafts correlates with reduced graft-versus-host disease.

BACKGROUND: Allogeneic hematopoietic cell transplantation (HCT) can be devastating when graft-versus-host disease (GvHD) develops. GvHD is characterized by mucosal inflammation due to breaching of epithelial barriers. Innate lymphoid cells (ILCs) are immune modulatory cells that are important in the maintenance of epithelial barriers, via their production of interleukin (IL)-22 and their T cell suppressive properties. After chemo- and radiotherapy, ILCs are depleted, and recovery after remission-induction therapy and after allogeneic HCT is slow and incomplete in a significant number of patients, which is associated with an increased risk to develop acute GvHD. OBJECTIVE: To investigate whether the presence of mature ILCs within G-CSF-mobilized HCT grafts is correlated with the development of acute GvHD after allogeneic HCT. STUDY DESIGN: We analyzed ILCs in a cohort of 36 patients who received allogeneic HCT for a hematologic malignancy, by flow-cytometric immune-phenotyping of prospectively collected, cryopreserved peripheral blood mononuclear cells (PBMCs) and donor-derived HCT grafts collected for the same patients. Biased analysis, with ILCs defined as CD3-lineage-CD45+CD127+CD161+ lymphocytes, was performed using FlowJo version 10 software. Unbiased analysis was done using FlowSOM, which uses a self-organizing map (SOM) with a minimal spanning tree (MST) to define and visualize different clusters present in the samples. RESULTS: Remission-induction therapy significantly depleted ILCs from the blood, and patients who had a relatively low percentage of ILCs before allogeneic HCT were significantly more prone to develop acute GvHD, confirming previous findings in a separate cohort. Allogeneic HCT grafts, which were all obtained from the blood of G-CSF-mobilized healthy donors, contained ILCs at a frequency very similar to the peripheral blood of healthy individuals. The ILC subset composition was also comparable to that of the blood of healthy individuals, with the exception of NKp44+ ILC3s, which were significantly more abundant in HCT grafts. The relative ILC content of the graft tended to correlate with ILC reconstitution after allogeneic HCT, suggesting that peripheral expansion of transplanted mature ILCs may contribute to early ILC reconstitution after allogeneic HCT. Patients who received a relatively ILC-poor HCT graft had a significantly increased risk to develop acute GvHD, compared with patients who received relatively ILC-rich allogeneic HCT grafts. Unbiased phenotypic analysis with the FlowSOM algorithm confirmed that allogeneic HCT grafts of patients who developed acute GvHD contained a lower frequency of ILCs that clustered in NKp44+ ILC3 signature groups. CONCLUSION: The presence of ILCs in allogeneic HCT grafts is associated with a reduced risk to develop acute GvHD. These data suggest that enhancement of ILC reconstitution of ILC3s in particular, for example via adoptive transfer of ILCs, may prevent acute GvHD and has the potential to improve outcome of allogeneic HCT recipients.

Computational flow cytometry as a diagnostic tool in suspected-myelodysplastic syndromes.

The diagnostic work-up of patients suspected for myelodysplastic syndromes is challenging and mainly relies on bone marrow morphology and cytogenetics. In this study, we developed and prospectively validated a fully computational tool for flow cytometry diagnostics in suspected-MDS. The computational diagnostic workflow consists of methods for pre-processing flow cytometry data, followed by a cell population detection method (FlowSOM) and a machine learning classifier (Random Forest). Based on a six tubes FC panel, the workflow obtained a 90% sensitivity and 93% specificity in an independent validation cohort. For practical advantages (e.g., reduced processing time and costs), a second computational diagnostic workflow was trained, solely based on the best performing single tube of the training cohort. This workflow obtained 97% sensitivity and 95% specificity in the prospective validation cohort. Both workflows outperformed the conventional, expert analyzed flow cytometry scores for diagnosis with respect to accuracy, objectivity and time investment (less than 2 min per patient).

Stabilization of Human Whole Blood Samples for Multicenter and Retrospective Immunophenotyping Studies.

Whole blood is often collected for large-scale immune monitoring studies to track changes in cell frequencies and responses using flow (FC) or mass cytometry (MC). In order to preserve sample composition and phenotype, blood samples should be analyzed within 24 h after bleeding, restricting the recruitment, analysis protocols, as well as biobanking. Herein, we have evaluated two whole blood preservation protocols that allow rapid sample processing and long-term stability. Two fixation buffers were used, Phosphoflow Fix and Lyse (BD) and Proteomic Stabilizer (PROT) to fix and freeze whole blood samples for up to 6 months. After analysis by an 8-plex panel by FC and a 26-plex panel by MC, manual gating of circulating leukocyte populations and cytokines was performed. Additionally, we tested the stability of a single sample over a 13-months period using 45 consecutive aliquots and a 34-plex panel by MC. We observed high correlation and low bias toward any cell population when comparing fresh and 6 months frozen blood with FC and MC. This correlation was confirmed by hierarchical clustering. Low coefficients of variation (CV) across studied time points indicate good sample preservation for up to 6 months. Cytokine detection stability was confirmed by low CVs, with some differences between fresh and fixed conditions. Thirteen months regular follow-up of PROT samples showed remarkable sample stability. Whole blood can be preserved for phenotyping and cytokine-response studies provided the careful selection of a compatible antibody panel. However, possible changes in cell morphology, differences in antibody affinity, and changes in cytokine-positive cell frequencies when compared to fresh blood should be considered. Our setting constitutes a valuable tool for multicentric and retrospective studies. © 2020 International Society for Advancement of Cytometry.

CytoNorm: A Normalization Algorithm for Cytometry Data.

High-dimensional flow cytometry has matured to a level that enables deep phenotyping of cellular systems at a clinical scale. The resulting high-content data sets allow characterizing the human immune system at unprecedented single cell resolution. However, the results are highly dependent on sample preparation and measurements might drift over time. While various controls exist for assessment and improvement of data quality in a single sample, the challenges of cross-sample normalization attempts have been limited to aligning marker distributions across subjects. These approaches, inspired by bulk genomics and proteomics assays, ignore the single-cell nature of the data and risk the removal of biologically relevant signals. This work proposes CytoNorm, a normalization algorithm to ensure internal consistency between clinical samples based on shared controls across various study batches. Data from the shared controls is used to learn the appropriate transformations for each batch (e.g., each analysis day). Importantly, some sources of technical variation are strongly influenced by the amount of protein expressed on specific cell types, requiring several population-specific transformations to normalize cells from a heterogeneous sample. To address this, our approach first identifies the overall cellular distribution using a clustering step, and calculates subset-specific transformations on the control samples by computing their quantile distributions and aligning them with splines. These transformations are then applied to all other clinical samples in the batch to remove the batch-specific variations. We evaluated the algorithm on a customized data set with two shared controls across batches. One control sample was used for calculation of the normalization transformations and the second control was used as a blinded test set and evaluated with Earth Mover's distance. Additional results are provided using two real-world clinical data sets. Overall, our method compared favorably to standard normalization procedures. The algorithm is implemented in the R package "CytoNorm" and available via the following link: www.github.com/saeyslab/CytoNorm © 2019 The Authors. Cytometry Part A published by Wiley Periodicals, Inc. on behalf of International Society for Advancement of Cytometry.

The ORMDL3 asthma susceptibility gene regulates systemic ceramide levels without altering key asthma features in mice.

BACKGROUND: Genome-wide association studies in asthma have repeatedly identified single nucleotide polymorphisms in the ORM (yeast)-like protein isoform 3 (ORMDL3) gene across different populations. Although the ORM homologues in yeast are well-known inhibitors of sphingolipid synthesis, it is still unclear whether and how mammalian ORMDL3 regulates sphingolipid metabolism and whether altered sphingolipid synthesis would be causally related to asthma risk. OBJECTIVE: We sought to examine the in vivo role of ORMDL3 in sphingolipid metabolism and allergic asthma. METHODS: Ormdl3-LacZ reporter mice, gene-deficient Ormdl3-/- mice, and overexpressing Ormdl3Tg/wt mice were exposed to physiologically relevant aeroallergens, such as house dust mite (HDM) or Alternaria alternata, to induce experimental asthma. Mass spectrometry-based sphingolipidomics were performed, and airway eosinophilia, TH2 cytokine production, immunoglobulin synthesis, airway remodeling, and bronchial hyperreactivity were measured. RESULTS: HDM challenge significantly increased levels of total sphingolipids in the lungs of HDM-sensitized mice compared with those in control mice. In Ormdl3Tg/wt mice the allergen-induced increase in lung ceramide levels was significantly reduced, whereas total sphingolipid levels were not affected. Conversely, in liver and serum, levels of total sphingolipids, including ceramides, were increased in Ormdl3-/- mice, whereas they were decreased in Ormdl3Tg/wt mice. This difference was independent of allergen exposure. Despite these changes, all features of asthma were identical between wild-type, Ormdl3Tg/wt, and Ormdl3-/- mice across several models of experimental asthma. CONCLUSION: ORMDL3 regulates systemic ceramide levels, but genetically interfering with Ormdl3 expression does not result in altered experimental asthma.

Myeloid cell heterogeneity in cancer: not a single cell alike.

Tumors of various histological origins show abundant infiltration of myeloid cells from early stages of disease progression. These cells have a profound impact on antitumor immunity and influence fundamental processes that underlie malignancy, including neoangiogenesis, sustained cancer cell proliferation, metastasis and therapy resistance. For these reasons, development of therapeutic approaches to deplete or reprogram myeloid cells in cancer is an emerging field of interest. However, knowledge about the heterogeneity of myeloid cells in tumors and their variability between patients and disease stages is still limited. In this review, we summarize the most recent advances in our understanding about how the phenotype of tumor-associated macrophages, monocytes, neutrophils, myeloid-derived suppressor cells and dendritic cells is dictated by their ontogeny, activation status and localization. We also outline major open questions that will only be resolved by applying high-dimensional single-cell technologies and systems biology approaches in the analysis of the tumor microenvironment.

FloReMi: Flow density survival regression using minimal feature redundancy.

Advances in flow cytometry bioinformatics have resulted in a wide variety of clustering, classification and visualization techniques. To objectively evaluate the performance of such methods, common benchmarks such as the FlowCAP initiative have proven to be of great value. In this work, we report on a novel method, FloReMi, which was developed to tackle the most recent FlowCAP IV challenge. This challenge was formulated as a survival modeling problem, where participants were expected to design a model to predict the time until progression to AIDS for HIV patients. It is known that variability in progression rate cannot be fully predicted by simple CD4(+) T cell counts. However, it is hypothesized that the immunopathogenesis established early in HIV already indicates the course of future disease. Adequately estimating the progression rate of HIV patients is crucial in their treatment. Using an automated pipeline to preprocess the data, and subsequently identify and select informative cell subsets, a survival regression method based on random survival forests was built, which obtained the best results of all submitted approaches to the FlowCAP IV challenge.

A benchmark for evaluation of algorithms for identification of cellular correlates of clinical outcomes.

The Flow Cytometry: Critical Assessment of Population Identification Methods (FlowCAP) challenges were established to compare the performance of computational methods for identifying cell populations in multidimensional flow cytometry data. Here we report the results of FlowCAP-IV where algorithms from seven different research groups predicted the time to progression to AIDS among a cohort of 384 HIV+ subjects, using antigen-stimulated peripheral blood mononuclear cell (PBMC) samples analyzed with a 14-color staining panel. Two approaches (FlowReMi.1 and flowDensity-flowType-RchyOptimyx) provided statistically significant predictive value in the blinded test set. Manual validation of submitted results indicated that unbiased analysis of single cell phenotypes could reveal unexpected cell types that correlated with outcomes of interest in high dimensional flow cytometry datasets.

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.

Myelodysplastic neoplasms dissected into indolent, leukaemic and unfavourable subtypes by computational clustering of haematopoietic stem and progenitor cells.

Myelodysplastic neoplasms (MDS) encompass haematological malignancies, which are characterised by dysplasia, ineffective haematopoiesis and the risk of progression towards acute myeloid leukaemia (AML). Myelodysplastic neoplasms are notorious for their heterogeneity: clinical outcomes range from a near-normal life expectancy to leukaemic transformation or premature death due to cytopenia. The Molecular International Prognostic Scoring System made progress in the dissection of MDS by clinical outcomes. To contribute to the risk stratification of MDS by immunophenotypic profiles, this study performed computational clustering of flow cytometry data of CD34+ cells in 67 MDS, 67 AML patients and 49 controls. Our data revealed heterogeneity also within the MDS-derived CD34+ compartment. In MDS, maintenance of lymphoid progenitors and megakaryocytic-erythroid progenitors predicted favourable outcomes, whereas expansion of granulocyte-monocyte progenitors increased the risk of leukaemic transformation. The proliferation of haematopoietic stem cells and common myeloid progenitors with downregulated CD44 expression, suggestive of impaired haematopoietic differentiation, characterised a distinct MDS subtype with a poor overall survival. This exploratory study demonstrates the prognostic value of known and previously unexplored CD34+ populations and suggests the feasibility of dissecting MDS into a more indolent, a leukaemic and another unfavourable subtype.

A robust pipeline for high-content, high-throughput immunophenotyping reveals age- and genetics-dependent changes in blood leukocytes.

High-dimensional flow cytometry is the gold standard to study the human immune system in large cohorts. However, large sample sizes increase inter-experimental variation because of technical and experimental inaccuracies introduced by batch variability. Our high-throughput sample processing pipeline in combination with 28-color flow cytometry focuses on increased throughput (192 samples/experiment) and high reproducibility. We implemented quality control checkpoints to reduce technical and experimental variation. Finally, we integrated FlowSOM clustering to facilitate automated data analysis and demonstrate the reproducibility of our pipeline in a study with 3,357 samples. We reveal age-associated immune dynamics in 2,300 individuals, signified by decreasing T and B cell subsets with age. In addition, by combining genetic analyses, our approach revealed unique immune signatures associated with a single nucleotide polymorphism (SNP) that abrogates CD45 isoform splicing. In summary, we provide a versatile and reliable high-throughput, flow cytometry-based pipeline for immune discovery and exploration in large cohorts.

Adjuvant dendritic cell-based immunotherapy after cytoreductive surgery and hyperthermic intraperitoneal chemotherapy in patients with malignant peritoneal mesothelioma: a phase II clinical trial.

BACKGROUND: Malignant peritoneal mesothelioma (MPM) is an aggressive malignancy with a poor prognosis. Cytoreductive surgery (CRS) and hyperthermic intraperitoneal chemotherapy (HIPEC) improves survival outcomes, but recurrence rates remain high. Dendritic cell-based immunotherapy (DCBI) showed promising results in patients with pleural mesothelioma. The primary aim of this trial was to determine feasibility of adjuvant DCBI after CRS-HIPEC. METHODS: This open-label, single-center, phase II clinical trial, performed in the Erasmus MC Cancer Institute Rotterdam, the Netherlands, included patients with epithelioid MPM. 4-6 weeks before CRS-HIPEC leukapheresis was performed. 8-10 weeks after surgery, DCBI was administered three times biweekly. Feasibility was defined as administration of at least three adjuvant vaccinations in 75% of patients. Comprehensive immune cell profiling was performed on peripheral blood samples prior to and during treatment. RESULTS: All patients who received CRS-HIPEC (n=16) were successfully treated with adjuvant DCBI. No severe toxicity related to DCBI was observed. Median progression-free survival (PFS) was 12 months (IQR 5-23) and median overall survival was not reached. DCBI was associated with increased proliferation of circulating natural killer cells and CD4+ T-helper (Th) cells. Co-stimulatory molecules, including ICOS, HLA-DR, and CD28 were upregulated predominantly on memory or proliferating Th-cells and minimally on CD8+ cytotoxic T-lymphocytes (CTLs) after treatment. However, an increase in CD8+ terminally differentiated effector memory (Temra) cells positively correlated with PFS, whereas co-expression of ICOS and Ki67 on CTLs trended towards a positive correlation. CONCLUSIONS: Adjuvant DCBI after CRS-HIPEC in patients with MPM was feasible and safe, and showed promising survival outcomes. DCBI had an immune modulatory effect on lymphoid cells and induced memory T-cell activation. Moreover, an increase of CD8+ Temra cells was more pronounced in patients with longer PFS. These data provide rationale for future combination treatment strategies. TRIAL REGISTRATION NUMBER: NTR7060; Dutch Trial Register (NTR).