The ability to devise safe and effective medicines that elicit a targeted immune response to cancer or infectious disease hinges on a more integrated understanding of the cell and molecular mechanisms governing systemic immunology. A more holistic perspective can be achieved through detailed studies of immune structure and function using preclinical model systems that permit examination of immune response in multiple immune organs.

Although state of the art technologies for single-cell data acquisition allow for the rapid collection of expansive datasets pertinent to the study of systemic immune response, a companion data analysis tool capable of fully automating statistical analysis and data visualization is lacking. By leveraging a set of powerful Python-based computational libraries against bulky single-cell datasets spanning time, tissue, experimental perturbation, and biological replicate, SYLARAS overcomes the labor-intensity of manual cell subset identification, statistical analysis among subgroups, and visual display of single-cell data at scale.

Immune response programs can induce proliferation, migration, and differentiation of the biologically-specialized immune cell subsets comprising lymphoid tissue architecture. Although the molecular mechanisms underlying these changes may not always be clear, the resultant redistribution in immune cell frequency and inter-cellular correlation structure can be readily quantified by immunophenotyping: a technique for delineating the proportions of cell types based on differential antigen expression. SYLARAS uses this concept to infer changes in specific immune cell lineages and their network-level architectures secondary to an immune stimulus. The ability of SYLARAS to generate and display interpretable systemic immune response data makes it a broadly useful preclinical research tool.

When immunoreactivity is reported in binary terms (e.g. CD_a⁺, CD_b^-, CD_c⁺), the number of possible immunophenotypes specified by a panel of M antibodies is 2^M. This exponential relationship makes comprehensive cell subset identification through serial gating cumbersome, time consuming, and, in many cases, non-commutative. SYLARAS addresses these drawbacks of manual gating through computation by programmatically assigning an immunophenotype "bar code" to each cell in the analysis.

In the SYLARAS procedure, the interface between background and foreground signal is defined for each immunomarker by curating a set of one dimensional gates (histogram gates). The signal intensity distribution of unlabeled/isotype labeled cells is super-imposed on the immunolabeled distributions to aid in objective gate placement by revealing how autofluorescence and/or isotype control antibody binding compares to observed data ( panel a ).

The data then undergo a linear data transformation to center the predetermined gate values at zero. This causes background signal intensities to become negative valued and an M-dimensional Boolean immunophenotype to be programmatically assigned to each cell in the dataset ( panel b ). For example, a cell assigned the Boolean vector [1, 0, 1, 1, 0, 0, 0, 1, 1 ,0, 0] might correspond to the immunophenotype: CD_a⁺, CD_b^-+, CD_c⁺, CD_d⁺, CD_e^-, CD_f^-, CD_g^-, CD_h⁺, CD_i⁺, CD_j^-, CD_k^-

In the case of a three immunomarker panel, the vectorization procedure can be conceptualized geometrically as the binning of cells across the 2³ (= 8) octants of a 3-dimensional cube ( panel c ). Extrapolating the principle into higher dimensions (i.e. more immunomarkers), the vectorization procedure generalizes as the binning of cells across the 2^M orthants of an M-dimensional hypercube.

(1) An immune stimulus (glioblastoma brain cancer shown here) or control agent is administered to cohorts of age-matched mice. (2) Lymphoid tissues are harvested from mice of each treatment group at various time points after the onset of immune stimulation. (3) Tissues are processed into single-cell suspensions and plated in a 96 well V-bottom plate. (4) Cells are immunolabeled with a cocktail of fluorophore-conjugated antibodies then stained with a fixable viability dye (FVD). (5) Data are acquired by 12-color high-throughput flow cytometry. (6) Raw data are spectrally deconvolved and selected for viable singlets. (7) Pre-processed data undergo a computer-assisted gating procedure (see the "Cell Subset Identification" tab above for details) prior to computational analysis with SYLARAS software. Click here to download our experimental protocol.

SYLARAS portrays longitudinal, multi-organ immunophenotyping data on a per cell type basis in a readily-interpretable dashboard layout. The example of polymorphonuclear (PMN) immune cells shown here was programmatically generated during a SYLARAS screen of the mouse immune response to syngneic, orthotopic glioblastoma(GBM) brain cancer. (a) brief alias; (b) lineage; (c) immunomarker signature indicating whether the immunophenotype corresponds to 1 of 14 major “landmark population”; (d) distribution of cells across 5 lymphoid tissues color-coded as in (h, i and j); (e) percentage of this cell type relative to all immune cells; (f) forward and side scatter (FSC/SSC); (g) Logicle-transformed, background-subtracted immunomarker signal intensity; (h and i) time and tissue-specific difference in mean percentage and log₂ fold-change between GBM-burdened and mock-engrafted animals (n=8 mice/group), asterisks denote one of three levels of statistical significance; (j) contribution of this cell type (in percent) in each tissue across the study’s 48 mice.