Web-based analysis of multiparameter cytometry datasets.

Abstract

e13687 Background: High-dimensional cytometry is pivotal in oncology research for detailed immune profiling. However, challenges in sample collection, processing, and data analysis hinder its broader application. Analyzing complex datasets encompassing hundreds of immune populations demands significant expertise and resources, as current analysis workflows are time-consuming and can lack reproducibility. There is a need for improved cytometry data analysis tools that facilitate efficient and accurate interpretation of these datasets, ensuring reliable and reproducible results. Methods: We developed TokuProfile, a web-based dashboard to streamline the interpretation of high-dimensional cytometry data. The platform provides intuitive navigation through immune lineages, enabling users to explore hierarchical relationships among cell populations. It features interactive gated analyses, allowing for dynamic examination of specific cell subsets. Group similarity visualization is facilitated using Principal Component Analysis (PCA), aiding in the identification of patterns across different sample groups. The dashboard also tracks immune changes across various endpoints, such as time, dose, or response, providing insights into temporal dynamics and treatment effects. Unsupervised clustering analysis is available to identify rare or unexpected immune cell populations, enhancing the discovery of novel biomarkers. The capability of TokuProfile was validated using a dataset of 460 samples from 257 melanoma patients treated at Massachusetts General Hospital, profiled with a 44-marker mass cytometry panel. Results: The TokuProfile dashboard has demonstrated its ability to process multimillion-data-point datasets efficiently, delivering clear visualizations and actionable insights. While traditional analysis of high-dimensional cytometry datasets typically require weeks or months, TokuProfile generated intuitive and interpretable results within minutes. Users can navigate complex immune hierarchies and identify key markers driving data patterns. The unsupervised analysis successfully detected immune cell populations that traditional methods might overlook. The dashboard supports direct data export, including CSV files with population frequencies, functional subset frequencies, and median marker expression values, as well as FCS and GatingML files. All plots within the dashboard can be downloaded for integration into presentations and reports. Conclusions: TokuProfile effectively addresses the challenges inherent in high-dimensional cytometry data analysis by providing a simple, reproducible data analysis tool. Its application can facilitate the discovery of novel biomarkers and accelerate the development of immunotherapies. Additionally, we make datasets freely available for non-commercial (i.e. academic) use through TokuProfile.