Fig. 6

Constructing an age predictor model to explore human aging process. Workflow summary of PHARE construction (A) and applied PHARE to age prediction based on scRNA-seq and bulk RNA-seq data (B). C Spearman correlation analysis between predicted and actual age in discovery scRNA-seq datasets from healthy people. D Box plot showing age difference was not different between different healthy scRNA-seq cohorts. E Spearman correlation analysis between predicted and actual age in validation scRNA-seq datasets from healthy people. F Spearman correlation analysis between predicted and actual age in scRNA-seq datasets from diseased people. G Box plot showing age difference was significantly different between different disease cohorts. H Box plot showing age difference was significantly different between different stages of SLE. I Box plots showing age difference were significantly different between different clinical groups. J Spearman correlation analysis between predicted and actual age in bulk RNA-seq datasets from healthy people. K Box plot showing age difference was significantly different between healthy and CAD samples based on bulk RNA-seq data (GSE180081), the P values are calculated with t-tests (two-sided). In figure C, D and F, the P values are calculated with kruskal.test. In Figure G, H and J, the P values are calculated with t-tests (two-sided)