Evaluation of tSNE and FlowSOM unsupervised analysis in mouse blood flow cytometry data after an inflammatory challenge

Traumatic brain injury involves a primary mechanical injury that is followed by a secondary<br/>inflammatory cascade. The inflammatory cascade in the CNS releases cytokines which are<br/>associated with leukocytosis and a systemic immune response. Acute changes to peripheral<br/>immune cell populations post-TBI include a 4.5-fold increase of neutrophils 3 hours post-injury,<br/>and 2.7-fold or higher increase of monocytes 24 hours post-injury. Flow Cytometry is a<br/>technique that integrates fluidics, optics, and electronics to characterize cells based on their light<br/>scatter and antigen expression via monoclonal antibodies conjugated to fluorochromes. Flow<br/>cytometry is a valuable tool in cell characterization however the standard technique for data<br/>analysis, manual gating, is associated with inefficiency, subjectivity, and irreproducibility.<br/>Unsupervised analysis that uses algorithms packaged as plug-ins for flow cytometry analysis<br/>software has been discussed as a solution to the limits of manual gating and as an alternative<br/>method of data visualization and exploration. This investigation evaluated the use of tSNE<br/>(dimensionality reduction algorithm) and FlowSOM (population clustering algorithm)<br/>unsupervised flow cytometry analysis of immune cell population changes in female mice that<br/>have been exposed to a LPS-induced systemic inflammatory challenge, results were compared to<br/>those of manual gating. Flow cytometry data was obtained from blood samples taken prior to and<br/>24 hours after LPS injection. Unsupervised analysis was able to identify populations of<br/>neutrophils and pro-inflammatory/anti-inflammatory monocytes, it also identified several more<br/>populations however further inquiry with a more specific fluorescent panel would be required to<br/>establish the specificity and validity of these populations. Unsupervised analysis with tSNE and<br/>FlowSOM demonstrated the efficient and intuitive nature of the technique, however it also<br/>illustrated the importance of the investigator in preparing data and modulating plug-in settings.