LAMBDA-R

LAMBDA-R performs model-based clustering of high-dimensional flow and mass cytometry data while integrating clinical information to jointly identify unknown cell populations and their associations with clinical outcomes.

Key Features:

• LAMBDA (Latent Allocation Model with Bayesian Data Analysis): performs simultaneous identification of cell populations and association analysis with clinical data using a unified statistical framework.
• Probabilistic modeling: specifies probabilistic models tailored to cytometry type, including a zero-inflated distribution for mass cytometry data.
• High-dimensional data handling: manages flow and mass cytometry single-cell data across over 10 protein profiles for systemic-level characterization.
• Statistical validation: a simulation study demonstrated accurate estimation of model parameters.
• Clinical integration: jointly analyzes cytometry measurements and clinical information to discover associations between cell populations and clinical outcomes.

Scientific Applications:

• Disease diagnosis: identification of cell-population signatures from flow and mass cytometry associated with disease states.
• Prognosis and predictive modeling: discovery of population-level biomarkers linked to clinical outcomes for use in predictive models.
• Translational research and personalized medicine: linking single-cell cytometry profiles with clinical information to support translational studies and individualized analyses.
• System-level biological characterization: characterization of complex biological systems at single-cell resolution using high-dimensional cytometry data.

Methodology:

The approach uses Bayesian data analysis via a latent allocation (LAMBDA) model for model-based clustering, employs probabilistic models including a zero-inflated distribution for mass cytometry, and is implemented in R.