Fig. 1: Transcriptome-based aging clocks can predict the ages of individual neuron types.
a, Distribution of transcriptomic age predictions. The 128 neurons of the CeNGEN dataset were predicted with BitAge (upper) or a stochastic noise-based clock (lower) and sorted by their predicted age. The x axis shows the rank of the prediction in ascending order, and the y axis shows the predicted age. The 20% youngest neurons and their respective age predictions are outlined in blue; the 20% oldest neurons and their respective age predictions are displayed in orange. Neuron types appearing on the top/bottom of both prediction lists are indicated in bold letters. b, Predictions with BitAge and the Stochastic Clock on the CeNGEN dataset are highly correlated (Pearson correlation 0.65, P value 5.5 × 10−17). The x axis shows the BitAge predictions of the CeNGEN dataset, and the y axis shows the stochastic data-based clock predictions of the CeNGEN dataset. All 128 neurons are plotted. Color coding of the youngest 20% of neurons according to BitAge in blue and of the oldest 20% in orange. The regression model fit with a 95% confidence interval (dashed, black lines) is shown. c, Neuron-type-specific genes in the CeNGEN dataset are skewed depending on gene length and age prediction. The log10 gene length (x axis) of the specific genes from the 20% youngest neurons (blue), and of the 20% oldest neurons (orange) are compared to random permutation of all neuron-type-specific genes with the same number of genes (gray). The gray curve shows the mean density across 100,000 random permutations, each sampling the same number of genes as in the youngest/oldest neuron groups; the shaded gray band indicates the 95% confidence interval across permutations. The two-sided permutation test compared the median log10 gene length. The y axis shows the probability density of the values on the x axis. d, Swarm plot of the median log10 length of neuron-type-specific genes expressed in the youngest 20% (blue) and oldest 20% (orange) neurons according to BitAge predictions. Each point represents a single neuron type. Neuron types were grouped into the youngest 20% (blue, n = 17) and oldest 20% (orange, n = 21) according to BitAge-predicted biological age. Data shown are the mean ± s.d. Welch’s two-tailed t-test was used to test for significant differences.
