Stability Dashboard
Evaluate epigenetic clock robustness to technical and biological confounders
Desktop Required
For analysis features, please use the desktop version of this page. The interactive dashboard requires a larger screen.
Filter Clocks
Use Case 1: Generation 2 Clock Confounders
Use Case 2: Systematic Clock Score Changes
Stability Analysis Methodology
Understanding our approach to measuring clock reliability and robustness
Mathematical Framework
ICC Calculation
For each epigenetic clock and dataset combination, ICCs are calculated using mixed-effects models:
Yij = μ + ui + εij
- Yij: Clock value for subject i at measurement j
- μ: Overall mean
- ui ~ N(0, σu²): Random effect for subject i
- εij ~ N(0, σe²): Measurement error
ICC = σu² / (σu² + σe²)
Meta-Analysis Approach
Fisher's z-transformation for approximate normality:
z = ½ ln((1 + ICC) / (1 - ICC))
Variance of transformed ICC:
Var(z) = 1 / (n - 3)
where n is the number of subjects.
Random-Effects Meta-Analysis
For each clock and replicate type combination using the metafor package in R:
zi = θ + ui + εi
- zi: Transformed ICC from study i
- θ: Overall effect size
- ui ~ N(0, τ²): Between-study heterogeneity
- εi ~ N(0, vi): Within-study variance
Back-transformation to ICC scale:
Pooled ICC = (exp(2θ̂) - 1) / (exp(2θ̂) + 1)
Confounders Assessed
Technical Confounders
- • Array batch effects
- • Laboratory processing variation
- • Storage conditions
- • DNA extraction methods
- • Bisulfite conversion efficiency
Biological Confounders
- • Meal timing and composition
- • Sleep duration and quality
- • Circadian rhythm effects
- • Stress and cortisol levels
- • Physical activity patterns
ICC Interpretation Guidelines
Based on Koo & Li (2016)
High sensitivity to confounders
Moderate sensitivity to confounders
Low sensitivity to confounders
Robust to confounders
Implementation Features
Key Features
- • Multiple timepoints per individual
- • Technical and biological replication
- • Standardized ICC calculation
- • Interactive confounder filtering
- • Comprehensive stability metrics
Clinical Implications
- • Clock selection for clinical use
- • Sample collection protocols
- • Quality control standards
- • Measurement error assessment
- • Longitudinal study design
Stability Implications for Clock Selection
High Stability Clocks
Suitable for:
- • Clinical diagnostics
- • Longitudinal monitoring
- • Multi-site studies
- • Real-world applications
Low Stability Clocks
Require:
- • Strict protocols
- • Controlled environments
- • Technical replication
- • Careful interpretation