censcyt

censcyt implements methods for differential abundance analysis in high-dimensional cytometry data when a covariate is subject to right censoring, using multiple imputation and generalized linear mixed models.

Key Features:

• Censored Covariate Handling: Enables direct inclusion of right-censored covariates (e.g., survival time) as predictors in differential abundance analysis.
• Multiple Imputation Methods: Provides multiple imputation-based approaches for analyzing censored covariates with performance comparable to Cox proportional hazards models.
• Generalized Linear Mixed Model Integration: Implements methods to incorporate censored variables into generalized linear mixed models (GLMMs) for flexible differential analysis.
• diffcyt Extension: Functions as an extension of the diffcyt workflow, integrating with its preprocessing, cell population identification, and differential testing framework.

Scientific Applications:

• Clinical Survival Studies: Analyzes cell composition changes in cytometry data where survival time is a censored covariate in clinical cohorts.
• High-Dimensional Cytometry Differential Analysis: Performs differential abundance testing for cell populations across conditions with incomplete follow-up data.
• Single-Cell State Transition Analysis: Supports investigation of changes in cell states and transitions in datasets with censored clinical variables.

Methodology:

The methods are based on multiple imputation and generalized linear mixed models, implemented as an extension of the diffcyt workflow for differential discovery in cytometry data.