CytoScan: Automated Detection of Technical Anomalies for Cytometry Quality Control

Studies evaluating cellular phenotypes by cytometry techniques are increasingly facing analytical challenges due to the multitudes of samples and parameters that are evaluated concurrently. Spurious technical effects resulting from a lack of standardization can affect marker distributions and further complicate multi-sample analyses. User-friendly tools for exploratory data analysis to identify such technical effects in large datasets are lacking. To fill this gap, we present a novel R package, CytoScan, that evaluates inter-measurement variation in cytometry datasets and allows for detecting anomalous measurements after data acquisition. CytoScan can detect two types of anomalies: files with limited similarity to others within a dataset (outliers) and files with limited similarity to previously acquired high-quality reference data (novelties). Using simulations of skewed marker distributions and real-life technical effects, we demonstrate that CytoScan can accurately detect such anomalies. CytoScan can be applied to large cytometry datasets on consumer-grade hardware with informative visualizations, providing accessible quality control for more reliable analyses.