CAcnesDB: A database of Cutibacterium acnes with Integrated Functional Insights derived from Multi-modal Genome Annotation

Cutibacterium acnes is a predominant member of the human skin microbiome, generally as a beneficial or a benign inhabitant, which under specific conditions transforms into a pathogen, causing acne and associated inflammatory responses. While there is a wealth of data on the microbiology and genomics of the microbe, there is a lack of comprehensive annotation of the proteins coded by the genome, slowing progress in understanding its unique metabolic versatility and survival in host hypoxic sebaceous niches. We configured an integrated pipeline for functional annotation of the C. acnes genome, combining gene calling, domain and ontology mapping, KEGG pathways assignment, proteome-scale structural modeling, predicting Pockets and associating the small-molecule metabolites based on binding site similarity to facilitate structure-to-function annotation. Structural predictions using ColabFold covered nearly the full proteome, with 67.4% of models showing high confidence. To infer protein function, ligand-binding sites in the C. acnes modeled proteome were predicted using a consensus of three independent algorithms, PocketDepth, SiteHound, and FPocket. The resulting pockets were matched with known small-molecule binding sites in PDB to map potential ligands, resulting in the C. acnes predicted metabolome (CApM) comprising 1,954 ligands across 1,865 proteins. CAcnesDB is a comprehensive up-to-date database of the C. acnes genome, hosting functional annotation from multiple levels of input and analyses, covering 2297 protein-coding genes.