What is UniProt?

UniProt is organized into three main components. Swiss-Prot contains manually annotated, reviewed protein entries — each curated by expert biologists who synthesize information from published literature, large-scale studies, and computational analyses. TrEMBL contains automatically annotated entries that have not yet been manually reviewed. UniRef provides clustered protein sequence sets at different identity thresholds (100%, 90%, 50%) for efficient similarity searches. For drug design and pharmacogenomics, Swiss-Prot is the gold standard because its annotations include detailed information on protein function, domain architecture, post-translational modifications, disease associations, and variant effects.

The power of UniProt for drug design lies in its feature annotations. Each protein entry includes the complete amino acid sequence, known functional domains (e.g., active sites, binding sites, transmembrane regions), experimentally characterized variants with their phenotypic effects, tissue expression patterns, protein-protein interactions, and cross-references to over 170 external databases including PDB (3D structures), Pfam (protein families), and Reactome (biological pathways). This interconnected annotation allows researchers to understand not just what a protein does, but how specific genetic variants alter its function at the molecular level.

PepFold uses UniProt as its primary protein knowledge base. After retrieving variant information from ClinVar, the pipeline queries UniProt to obtain the full protein sequence, identify the functional domain affected by the variant, retrieve known 3D structural information, and map the variant's position relative to active sites, binding pockets, and protein-protein interaction interfaces. This structural and functional context drives the peptide design algorithm: candidates are generated to target specific protein domains with full awareness of how the patient's genetic variant alters the target protein's structure and function.