Codes for reproducibility of the paper "Hidden Links between Skin Microbiome and Skin Imaging Phenome"
Despite skin microbiome has been linked to skin health and diseases, its role in modulating human skin appearance remains understudied. Although a few studies have identified the microbial correlations with specific skin appearance features, the majority of them remained underexplored due to the lack of the large-scale cohorts with comprehensive omics data. Using a total of 1,244 face imaging phenomes and 246 cheek metagenomes, we firstly established three skin age indices by machine learning including skin phenotype age (SPA), skin microbiota age (SMA), and skin integration age (SIA) as surrogates of phenotypic aging, microbial aging, and their combination, respectively. Moreover, we found that besides aging and gender as intrinsic factors, skin microbiome also played a role in shaping skin imaging phenotypes (SIPs). Skin taxonomic and functional α diversity was positively linked to melanin, pore, spot, and ultraviolet spot levels, but negatively linked to sebum, lightening, and porphyrins levels. Furthermore, certain species were correlated with specific SIPs, such as sebum and lightening levels negatively correlated with Corynebacterium matruchotii, Staphylococcus capitis, and Streptococcus sanguinis. Notably, we demonstrated skin microbial potential in predicting SIPs, among which the lightening level presented the least error of 1.8%. Lastly, we provided a reservoir of potential mechanisms through which skin microbiome adjusted the SIPs, including the modulation of pore, wrinkle, and sebum levels by cobalamin and heme synthesis pathways, predominantly driven by Cutibacterium acnes. This pioneering study unveils the paradigm for the hidden links between skin microbiome and skin imaging phenome, providing novel insights into how skin microbiome shapes skin appearance and its healthy aging.
The directory "data/" stores the input files for the codes, including skin imaging phenome matrices and skin metagenomic composition matrices.
The directory "data/Figure3_input" stores the gene composition matrices and functional composition matrices, as well as their annotations.
The directory "intermediate_results/" stores the intermediate results produced by the codes under the directory "scripts/".
The directory "figures/" stores the raw figures produced by the codes under the directory "scripts/".
The directory "scripts/" stores all the codes for producing the main results and figures of the paper.
To use them, simply download the whole package and run in the R.