4. Project Types for Computational Protein Design

Ariax Bio supports two engines for computational binder design:

• BindCraft — automated de novo design for miniproteins and linear peptides, with robust filtering and strong few‑shot performance. For scoring, choose between PyRosetta and fully open‑source FreeBindCraft (no Rosetta license). Learn more about why BindCraft is revolutionizing protein design and our FreeBindCraft announcement.
• BoltzGen — an all‑atom generative model for universal binder design across modalities. Use it for VHH (nanobody), cyclic‑peptide, helicon, miniprotein, peptide, and miniprotein + small‑molecule targeting. See BoltzGen Project Setup and our news post.

Before starting your project:
Use our Prep Inputs tool to trim your target structure, visualize molecular features, and select hotspots—all in your browser, completely free. See the complete guide for detailed instructions.

4.1 Miniprotein

De novo miniprotein design creates small proteins—typically 60-200 amino acids (below 15 kDa in molecular weight)—that are designed entirely from scratch using computational methods. Unlike traditional protein engineering, which modifies existing natural proteins, de novo design creates new amino acid sequences and structures not found in nature.

Miniproteins are notable for their compact size and their ability to form stable, well-defined three-dimensional structures. Their small size makes them easier to produce and highly stable, which is advantageous for both research and therapeutic applications. Because of these properties, de novo miniproteins are highly modular and can be tailored for specific structural or functional goals, such as binding to a particular target.

The benefits of de novo miniproteins are wide-ranging. They hold significant therapeutic potential, serving as agents for various diseases, including use as recombinant biologics, components of engineered cell therapies, lipid nanoparticles, and viral gene delivery platforms. Additionally, these proteins can act as building blocks for new biomaterials and, owing to their specificity, are valuable reagents for a variety of applications in biochemistry, molecular biology, and cell biology.

4.2 VHH (Nanobody)

Single‑domain antibodies (~15 kDa) derived from camelid heavy‑chain–only antibodies. Their compact footprints and elongated CDR3 loops allow VHHs to access recessed or cryptic epitopes that are difficult for conventional IgGs to reach. They are typically highly soluble and thermally stable.

VHHs are straightforward to express in microbial hosts and can be engineered into a wide variety of therapeutic formats, including multivalent or multispecific fusions, Fc‑extended constructs for half‑life, and payload‑bearing biologics. Their small size supports deep tissue penetration and makes them attractive as targeting domains for cell therapies and modular biologic architectures.

Key considerations include a shorter serum half‑life in the absence of extension strategies (e.g., Fc or albumin fusion, PEGylation), and lower monovalent affinity relative to full IgGs where avidity or maturation may be beneficial. Like other biologics, VHHs may require humanization to minimize immunogenicity depending on the use case.

4.3 Peptide

Short amino‑acid chains (typically 8–30 residues) that bridge the space between small molecules and proteins. Peptides present broad interaction surfaces and can engage protein–protein interfaces that are often intractable to traditional small‑molecule chemotypes.

They are fast to synthesize, highly tunable through sequence and chemistry, and offer excellent specificity with limited off‑target pharmacology. As such, peptides are valuable as research probes, as targeted modulators, and as starting points for lead optimization or scaffold elaboration.

Their main challenges are protease susceptibility and rapid renal clearance. Practical peptide therapeutics often employ cyclization, stapling, backbone modification, or non‑canonical residues to improve stability and permeability, and may use carriers or depot strategies to extend exposure.

4.4 Cyclic‑Peptide

Macrocyclized peptides (typically 8–30 aa) in which termini or side chains are covalently linked to enforce compact conformations. Conformational restriction can improve protease resistance, reduce entropic penalties on binding, and, in some scaffolds, enhance membrane permeability.

Cyclic peptides often show better stability and pharmacokinetics than their linear analogs and can achieve higher potency or selectivity through pre‑organization. With permeability engineered appropriately, some scaffolds are suited to intracellular targets and challenging interfaces.

Trade‑offs include more complex synthesis and structure–activity relationships arising from macrocyclization chemistry. Permeability and oral bioavailability remain dependent on ring size, polarity, and scaffold design, and must be balanced against potency and solubility.

4.5 Helicon

Stapled α‑helical peptides that use covalent “staples” (e.g., hydrocarbon or lactam linkages) to stabilize helical structure and present a continuous binding face. Helicons are designed for protein–protein interfaces dominated by helical motifs and can mimic natural interaction surfaces.

Stabilizing the helix often improves protease resistance and can increase affinity by preserving the bioactive conformation. With appropriate physicochemical tuning and staple geometry, helicons can exhibit cellular uptake and engage cytosolic targets, aided by non‑canonical residues and precise staple positioning.

Practical constraints include specialized synthesis and purification—often at higher cost than linear peptides—and an inherent bias toward targets with helical binding grooves. Permeability, solubility, and potency must be jointly optimized for each scaffold and indication.

4.6 When to Use Which Engine

• Choose BindCraft for miniproteins and linear peptides when all‑atom generative constraints are not required. BindCraft is highly automated, battle‑tested, and fast to evaluate with strong filters.
• Choose BoltzGen when your therapeutic concept needs:
  • Antibodies (VHH), cyclic‑peptides, or helicons
  • All‑atom reasoning or a rich binding‑site constraint language
  • Small‑molecule targeting (CCD/SMILES)

Is there a protein design tool you would like to see hosted at Ariax Bio? Let us know at info@ariax.bio.