Extended Data Fig. 1: Site parameters derived from SCARLET across CpG categories.
From: A unifying model of stem cell dynamics explains age-related methylation patterns across mammals
a, Comparisons of model fits across site categories, across varying sample sizes and numbers of selected sites. Models compared using expected log pointwise predictive density using leave-one-out cross-validation. Points represent means and error bars indicate standard deviations. Estimates use the approximately optimal value of \(N/s\)=10,000, while \(s\) is fixed at a plausible value of 1. b, The absolute difference between the inferred starting methylation value (at 0 years post sexual maturity, p) and steady-state methylation level approached as age increases (η) against the sum of epigenetic error probabilities, \({\,P}_{M\to U}\) and \({\,P}_{U\to M}\). Estimates use the approximately optimal value of \(N/s\)=10,000, while \(s\) is fixed at a plausible value of 1. c, Boxplots (showing median and interquartile range) of epigenetic maintenance efficiencies when \(N/s\) is the optimal value shown in Fig. 2b (10,000) and the rate of stem cell division, \(s\), is fixed at a realistic value of 1 division per stem cell per year. Sites are split into those that increase or decrease in mean methylation with age (note that for VMPs, this categorization will be essentially arbitrary). Estimates taken from 200 sites and 200 samples for each category.
