The glmmPen R package grew out of Rashid Lab collaborations with UNC Lineberger investigators who needed principled inference for clustered, high-dimensional datasets—ranging from bulk RNA-seq to clinic-based registries and survival studies. The software implements penalized generalized linear mixed models and mixed-effects Cox models with automated tuning and diagnostic utilities.

Why it matters

• Precision modeling. Handles thousands of fixed and random effects simultaneously, empowering biomarker discovery and risk prediction studies with repeated measures and time-to-event outcomes.
• Survival analysis integration. Extends penalized mixed models to Cox proportional hazards regression, enabling discovery of prognostic signatures in clustered cancer datasets with frailty terms.
• Production ready. Exposes user-friendly functions for simulation, cross-validation, and reporting, and underpins analyses in ARPA-H ADAPT, SPORE projects, and industry partnerships.

Key publications

• Penalized GLMMs: (Heiling et al., 2024)
• Survival extensions: Variable selection in high-dimensional mixed effects survival models (Biometrics, 2024) - arXiv:2504.00755