flowVS

flowVS stabilizes variance in flow cytometry (FC) and microarray fluorescence data by removing mean-variance correlations to enable accurate comparison of phenotypes across heterogeneous cell populations.

Key Features:

• Variance Stabilization Algorithm: Applies an inverse hyperbolic sine (asinh) transformation to fluorescence channels across all samples to remove mean-variance correlations.
• Parameter Optimization via Bartlett's Test: Selects asinh transformation parameters using Bartlett's likelihood-ratio test to achieve homogeneous variances (homoskedasticity) across channels.
• Cell Population Identification: Identifies cell populations within each fluorescence channel prior to transformation to target within-population variance stabilization.
• Applicability to Microarray Data: Extends the asinh-plus-Bartlett approach to microarray datasets and performs comparably to VSN for microarray variance stabilization.
• Comparative Performance: Empirical evaluations on two publicly available FC datasets show more even within-population variance stabilization compared to flowTrans, flowScape, logicle, and FCSTrans.

Scientific Applications:

• Phenotypic Comparisons: Enables uniform feature extraction and comparison of phenotypically identical cell populations across samples, facilitating analyses in immunology and oncology.
• Data Normalization: Stabilizes variance across diverse experimental conditions to improve the reliability of downstream analyses.

Methodology:

Identifies cell populations per fluorescence channel and applies an asinh transformation with parameters optimized by Bartlett's likelihood-ratio test to enforce homoskedasticity across clusters.