Fig. 4

Outlier samples are more prevalent in tumor-adjacent samples and exhibit accelerated aging. a A table indicating the number of outliers and not-outliers in normal adjacent and normal samples. The proportion of samples in each group is represented by percentages, and significance was tested by the Fisher exact test (p value = 0.0001, OR=3.9, 95% CI 1.8–8.1). b The absolute difference between the predicted and the chronological age on the y-axis for not-outlier (gray) and outlier (red) samples in normal samples and normal adjacent to cancer samples. The numbers above the box plots represent the number of samples in each group. Significance was calculated by the Wilcoxon test. c The outliers are represented in three groups in both normal and normal adjacent to cancer tissues. The numbers on the bar graphs represent the number of samples in each group. In red are outliers with an age difference of increase of 10 years older or more between the predicted age and the chronological age, in blue are outliers with an age difference of 10 years younger or lower, and in gray are outliers with an age difference between 10 years younger and 10 years older. d Overlap of outlier patient samples identified by 146 age-related sites and outliers identified by random sites in 95% of the permutations. The significance of overlap was tested by the hypergeometric test (p value = 2.23 × 10⁻¹⁶). e Prediction evaluation of the difference between the chronological age and the predicted age in outliers defined by random sites or by aging sites. The x-axis is the mean absolute error (MAE) of the difference between the chronological age and the predicted age for each sample in each of the 1000 randomly generated outlier detection models. The red density plot represents the distribution of MAE of outliers from the 1000 predicted models; the gray density plot is the distribution of the not-outlier from the same 1000 predictions. The red (outlier) and gray (not-outlier) dashed lines represent the MAE values based on the aging sites