flowMagic gating benchmark: automated and manual cell population annotation for flow cytometry data analysis

This dataset provides a comprehensive benchmark for evaluating flow cytometry gating algorithms. This includes the flowMagic tool, which automates cell population identification without requiring a template from the input panel. The dataset includes trained models, annotated datasets and performance metrics comparing flowMagic against other gating tools such as FlowSOM and Elastigate. The dataset covers both template-based and generalized models trained on over 9,000 manually gated bivariate plots derived from multiple experimental panels. These manual annotations were obtained through Project Discovery, a citizen science initiative integrated into the online game "EVE Online". The dataset also includes COVID-19 panels, training and testing scripts in R, model files (for consensus gates and non-consensus gates training), F1 score evaluations, gating difficulty metrics and manually sorted visual validations performed by independent researchers. Each gating result is structured to reflect bivariate expression data and corresponding gate labels. Reference gates, model outputs and evaluation results are stored in CSV, RData and image formats. This dataset supports method development, reproducibility assessments and cross-platform benchmarking for automated gating algorithms in single-cell cytometry.