Cyto-Feature Engineering

Cyto-Feature Engineering identifies and characterizes immune cell populations from high-dimensional flow cytometry data to enable analysis of their associations with disease.

Key Features:

• High-Dimensional Data Handling: Processes datasets generated by modern flow cytometers with up to 50 parameters per cell and millions of cells per sample.
• Feature Engineering with Immunological Context: Applies feature engineering informed by immunological context and generates plots to aid interpretation of complex marker patterns.
• Threshold Development Using FMO Controls: Uses Fluorescence Minus One (FMO) controls or distinct population differences to establish thresholds and convert continuous marker intensities into binary positive/negative calls.
• Data Filtering and Refinement: Filters identified cell phenotypes to refine results to populations of interest within immune lineages.
• Statistical Correlation and Modularity: Performs statistical correlation of partitioned cytometry data with other experimental measurements and supports modular substitution of statistical tests, alternative initial gating steps, and integration with additional datasets.
• Validation and Versatility: Validated by comparison to manual gating of murine splenocytes and human whole blood and applicable across different cytometer types and panel designs.

Scientific Applications:

• Immune population characterization: Identification and characterization of immune cell subsets, including rare populations, in murine and human samples.
• Disease association analysis: Correlation of cell population partitions with experimental measurements to study associations with disease.
• High-dimensional cytometry studies: Analysis of large-scale, multiparameter flow cytometry datasets (up to 50 parameters, millions of cells) across diverse panels and instruments.
• Cross-dataset integration and comparative studies: Integration and comparison of partitioned cytometry data with other experimental datasets for multi-modal analyses.

Methodology:

Implemented in R; transforms continuous marker intensities to binary calls using Fluorescence Minus One (FMO) controls or distinct population differences; applies feature engineering with immunological context and produces plots; filters identified phenotypes and performs statistical correlation of partitioned data with other experimental measurements; modular design allows customization of statistical tests, alternative initial gating steps, and integration with additional datasets; validated by comparison to manual gating of murine splenocytes and human whole blood.