Towards automated gating of clinical flow cytometry data

Abstract

Flow cytometry analysis is widespread practice in cell biology, immunology and hematology. Cell populations of interest are typically identified by consecutively examining the expression levels of antigen marker pairs. Since this manual gating process lacks standardization and is time-consuming, several machine learning (ML) methods for automated gating of flow cytometry data have been proposed in recent years. However, their translation into routine workflows has been limited. To address this, we developed the Python package FLAG-X ("flow cytometry automated gating toolbox"), which supports two novel workflows that integrate manual with ML-based gating, using labeled and unlabeled training data. We selected state-of-the-art ML methods developed for automated gating for inclusion in FLAG-X, based on their gating performance in comparison to manual expert annotations. FLAG-X provides a unified interface for top-performing methods and enables seamless integration with standard software for manual gating by exporting results as FCS files. To demonstrate its practical utility, we applied FLAG-X to representative cases from clinical practice. FLAG-X is available at https://anaconda.org/channels/bioconda/packages/flagx/overview. Author summaryIn our research, we work with flow cytometry data, a common laboratory technique to measure the expression of specific antigens in individual cells of a patient sample. In everyday clinical diagnostics or research, experts use dedicated software tools to "manually" inspect these data and often select a subset of cells for further analysis, e.g. B cells for Lymphoma subtype classification. However, this manual procedure, known as gating, can be time-consuming and is not standardized such that outcomes may vary based on the clinician carrying out the cell selection. Automated, machine learning-based methods developed to mitigate those problems are rarely used in routine clinical work. In this study, we set out to close this gap by developing novel workflows that allow clinicians to more easily integrate (semi-)automated approaches into their existing workflows. To do this, we we developed a software package that brings together top-performing methods, functionality to handle clinical flow cytometry data and is designed to be compatible with with existing tools for manual analysis. We evaluated our workflows and toolbox using realistic clinical scenarios and considered practical requirements for their use. By integrating automation into familiar workflows, we strive to make flow cytometry gating both faster and more consistent to support wider use of computational methods in clinical diagnostics.