We provide the first example of how these recently developed standardized file formats can be used to report FCM data including analysis and results in a standardized, computationally exchangeable form within an ACS container (see Additional file 1). The ACS container captures data and an analytical pipeline as shown in Figure 1. The ACS container includes two FCS data files acquired on an LSR II cytometer as part of an immune response study, two preprocessed FCS data files, two Gating-ML files capturing the process of manual gating, two CLR files with results of automated clustering, and an XML-based table of contents capturing relations among all the components. The original FCS files are randomly chosen from a set of FCM data samples of peripheral blood mononuclear cells (PBMC). Each sample is stimulated by a specific microbe-derived danger signal at a specific concentration in order to investigate in cellular detail the recognition and responses of human antigen presenting cells (APCs) to the danger signals of interest. APCs form a major group of cells in the immune system responsible for detecting foreign microbes and presenting them to other immunity cells for further actions. Subpopulations of APCs include monocytes, plasmacytoid dendritic cells (pDC), and myeloid dendritic cells (mDC), each of which differ in functionality as they respond differently when challenged with the same danger signals. APCs respond to danger signals in a number of ways, one of which is through secretion of cytokines. Cytokines are cell-signalling molecules that are used in intercellular communications, and can activate other immune cells to take further actions against microbes and pathogens.
Our example FCS data files contain measurements corresponding to physical characteristics of the cells (i.e., forward and side scatter), cell surface proteins (CD123, CD11c, MHC-II, CD14), and intracellular cytokines (IL-6, IL-12, TNF-α, IFN-α). During the analysis, an initial series of preprocessing steps were applied to the data. This included the removal of dead cells and debris, compensation and subsetting to the desired channels. Then the cell surface proteins measurements of the preprocessed FCS files are used to gate the files manually with the purpose of identifying cell populations including monocytes, B cells, and dendritic cells (DCs, including myeloid and plasmacytoid subpopulations - mDCs and pDCs), as well as two additional novel populations, whose roles in the immune response still need to be further investigated. Parallel to the manual approach, the preprocessed FCS files were analyzed using the SamSPECTRAL [12] clustering algorithm, followed by a semi-automated cluster matching technique to identify the desired populations automatically. Populations identified by automated clustering were compared to manually gated populations in order to determine the suitability of this automated approach in identification of cell populations of interest. Information corresponding to the results of these analyses were captured in a computer processable form in the ACS container.
A combination of tools was used to create the ACS container and it's components. Compensation and the live cell gates were created in FlowJo. The spillover matrix and gate coordinates were exported from FlowJo and used in the flowCore [6] R/BioConductor [7] package in order to preprocess the FCS data files. The preprocessed FCS files were manually gated in FlowJo and the gates were exported in Gating-ML. In parallel, the SamSPECTRAL [12] clustering algorithm was applied on the preprocessed data and resulting clusters were labelled and the results were saved in the CLR file format using the write.table R function. All files were manually organized in file system folders in an ACS-like structure and a table of contents was created according to the ACS specification. The oXygen software tool (Syncro Soft Ltd.) was used to adjust and validate XML-based components of the container (i.e., the table of contents and the Gating-ML files). Finally, the ACS-like folder structure was zipped using Ubuntu's built-in compression support (any ZIP-compatible tool could have been used) and the file extension of the resulting file was changed to .acs as required by the ACS specification.