CODAK

CODAK applies a kernel distance covariance (KDC) multivariate statistical framework to test associations between compositional cell-type abundance data from mass cytometry and categorical or continuous predictors such as disease status.

Key Features:

• Kernel Distance Covariance (KDC): Uses the KDC framework to perform multivariate association testing on compositional data via kernel-based distance covariance measures.
• Compositional data handling: Explicitly addresses the non-Euclidean structure of cell type abundance data derived from mass cytometry.
• Association with categorical and continuous predictors: Tests associations between cell-type composition and predictors that are categorical or continuous, including disease status.
• Scalability and small-sample performance: Designed for high-dimensional mass cytometry datasets and maintains performance in small sample sizes (n < 25).
• Validation on simulated and real datasets: Evaluated through simulation studies and applied to real-world high-dimensional datasets, including subgroup comparisons such as Systemic Lupus Erythematosus (SLE) versus healthy controls.
• Implementation: Implemented in R.

Scientific Applications:

• Mass cytometry compositional analysis: Testing associations between cell-type abundance compositional profiles and biological or clinical predictors in mass cytometry studies.
• Small-sample, high-dimensional studies: Enabling association analyses when sample sizes are limited and feature dimensionality is large.
• Immunophenotyping and disease comparison: Comparing immune cell composition between clinical groups, for example Systemic Lupus Erythematosus (SLE) patients and healthy controls.

Methodology:

Performs multivariate association testing using the kernel distance covariance (KDC) framework on compositional cell-type abundance data, validated via simulation studies and applied analyses, and implemented in R.