What is De Novo Peptide Design?

Traditional peptide drug discovery starts with a known bioactive peptide — a hormone, a venom component, or a fragment of a natural protein — and modifies it to improve its drug-like properties. De novo design reverses this approach: starting from a target protein structure and desired binding mode, computational algorithms generate entirely new sequences that are predicted to fold into structures complementary to the target. Methods include physics-based approaches (Rosetta, molecular dynamics), evolutionary algorithms (genetic algorithms that 'evolve' peptide populations in silico), and increasingly, deep learning models trained on protein-peptide interaction data. The 2024 Nobel Prize in Chemistry, awarded to David Baker for computational protein design, validated this approach at the highest level.

The practical advantages of de novo design are substantial. Natural peptides often have suboptimal drug properties: short half-life, poor oral bioavailability, immunogenicity, or off-target binding. A de novo designed peptide starts with a blank slate and can be optimized simultaneously for multiple objectives. The sequence can incorporate non-natural amino acids (D-amino acids, beta-amino acids, N-methylated residues) that resist protease degradation. The backbone can be designed to adopt a specific conformation that maximizes target affinity. The overall physicochemical properties (charge, hydrophobicity, solubility) can be tuned for a specific administration route. This multi-objective optimization is difficult to achieve through modification of natural peptides alone.

PepFold's core technology is de novo peptide design driven by pharmacogenomic data. The pipeline starts with a patient's genetic variants, maps them to structural changes in target proteins via UniProt and ESMFold, and then generates novel peptide sequences computationally optimized to bind the variant-specific protein conformation. Each candidate is evaluated across 10 dimensions — binding affinity, structural confidence (pLDDT), protease resistance, ADMET properties, synthesis feasibility, aggregation propensity, permeability, novelty, clinical relevance, and predicted half-life. The top-ranked candidates are delivered with complete Fmoc-SPPS synthesis protocols, ready for laboratory validation.