Litepaper

DS-100 sAFGP series: how AI designs the next generation of synthetic antifreeze polypeptides

DS-100 is DeepSnow's R&D-stage polypeptide chemistry — alanine/glutamate alternating, NCA polymerization, 91–94% MGS reduction at 100 µg/mL. The discovery engine is what produces it.

DS-100 sAFGP series: how AI designs the next generation of synthetic antifreeze polypeptides

TL;DR. DS-100 is the R&D-stage product line in DeepSnow's pipeline: a series of synthetic antifreeze glycoprotein polypeptides (sAFGPs) with an alanine/glutamate-alternating backbone, produced by NCA polymerization. Lab assays show 91–94% mean grain-size reduction in ice recrystallization at 100 µg/mL — substantially higher IRI potency at lower dose than SL6733's polyacrylamide chemistry. The series is designed by DeepSnow's AI polymer discovery engine, engineered around existing intellectual property in the antifreeze-protein space. Pilot timeline: 2027/28+.

Why sAFGPs

Antifreeze glycoproteins (AFGPs) evolved in polar fish to prevent blood plasma from freezing in sub-zero seawater. They are the most potent natural ice-recrystallization inhibitors known: at micromolar concentrations they completely suppress ice crystal growth, far outperforming any synthetic polymer of comparable size.

The natural AFGPs are difficult to scale. They are glycoproteins (each repeat unit has a sugar moiety attached) and the natural production routes — extraction from cold-water fish — are not viable for industrial volumes.

Synthetic antifreeze glycoprotein polypeptides (sAFGPs) are the engineered alternative: polypeptides that capture the IRI-active properties of natural AFGPs without requiring glycosylation, biological extraction, or refrigeration to remain stable. They can be made at scale via standard polymer-chemistry routes.

This is the chemistry behind DeepSnow's DS-100 series.

The chemistry

DS-100 polypeptides have an alanine/glutamate alternating backbone — Ala-Glu-Ala-Glu-... — produced by N-carboxyanhydride (NCA) polymerization. The choice of monomers and the alternating sequence are deliberate:

  • Alanine is a small, hydrophobic amino acid with a methyl side chain. It contributes to the polymer's α-helical conformation under aqueous conditions and provides the "ice-binding face" geometry.
  • Glutamate has a carboxylate (COO⁻) side chain. It provides the active ice-surface binding sites — analogous to the carboxylate groups in SL6733's polyacrylamide, but at higher density and with controlled spatial periodicity.
  • NCA polymerization is a ring-opening polymerization of N-carboxyanhydride monomers — the standard method for producing synthetic polypeptides at controlled molecular weights and defined sequences.

The alternating Ala/Glu pattern is what produces the IRI activity. The amphipathic α-helical conformation displays one face of carboxylate groups (the ice-binding face) and one face of methyl groups (the hydrophobic face), in a geometry that matches the ice lattice spacing.

This design choice — and the broader sequence space around it — is what the DeepSnow discovery engine explores.

What the discovery engine does

The DeepSnow AI polymer discovery engine is a software system that ranks candidate polymer architectures by three predicted properties:

  1. Ice-binding affinity — how strongly does a given sequence bind the ice surface, predicted from molecular dynamics + sequence-conformation models?
  2. IRI potency — what is the expected mean-grain-size reduction at a target concentration?
  3. Manufacturability — can the candidate be synthesized at scale via NCA polymerization? Is the sequence stable, soluble, and producible?

A fourth constraint is IP-awareness: candidates are filtered against existing intellectual property in the antifreeze-protein space, ensuring DeepSnow's pipeline is engineered around prior art rather than into it.

The engine proposes candidate sequences. The wet lab synthesizes them. Splat assays + freeze-thaw characterization measure the actual IRI performance. Results feed back to the engine, which refines its model. This closed-loop iteration is what differentiates a discovery platform from a one-shot R&D effort.

The 91–94% MGS reduction number

In standard splat-assay protocols (a thin film of micro-crystals annealed at −8 °C for 30 minutes), DS-100 candidates at 100 µg/mL show:

  • Mean grain size (MGS) growth in untreated controls: roughly 4× over the annealing window.
  • MGS growth in DS-100-treated samples: 0.2–0.3× over the same window.
  • MGS reduction = 91–94% at the treated concentration vs control.

This is a substantial IRI potency. For comparison, SL6733's polyacrylamide chemistry at similar mass-based concentrations achieves roughly 50–70% MGS reduction. The DS-100 series operates at lower effective dose and produces denser, finer crystal structures.

Manufacturing pathway

NCA polymerization is industrially established. Companies producing synthetic polypeptides for pharmaceutical and biomedical applications — vaccine adjuvants, drug-delivery vehicles, biomaterials — operate at multi-ton scale. The chemistry, the safety profile, and the supply chain are mature.

DeepSnow's manufacturing strategy for DS-100 follows the SL6733 template:

  1. In-house wet-lab synthesis at pilot scale (sub-100L batches) for early lab pilots.
  2. Contract-manufacturing partnerships with established peptide manufacturers for commercial scale.
  3. Manufacturing process designed for the same drop-in operational profile as SL6733 — a polymer concentrate dosed at low ppm into resort water.

Timeline and dosing

  • Now (R&D): candidate generation and synthesis, splat assays, freeze-thaw characterization, manufacturing pathway development.
  • 2027/28 season: pilot trials at EU resorts (first cohort).
  • 2028/29+: commercial rollout.

The operational dose for DS-100 is expected to be substantially lower than SL6733 — single-digit ppm or lower — because of the higher per-molecule IRI potency. The precise commercial dose will be set after the pilot phase.

What DS-100 means for the platform

DS-100 is the first product where the value of DeepSnow's discovery engine is fully visible. SL6733 is a polymer-chemistry product engineered against a well-understood mechanism (high-MW polyacrylamide IRI). DS-100 is a product where the specific molecular sequences are determined by the engine's exploration of a vast design space — far larger than what a human chemist could systematically explore.

This is what makes DeepSnow a platform rather than a single-product company. The same wet lab + AI engine that produced DS-100 produces DS-400 (ice rink IRI), and produces the candidates for the future verticals: food-IRI for cold-chain food handling, cryopreservation for biologics, surface IRI for cold-chain logistics.

What to read next

DS-100 is in R&D. Public technical content reflects engineering targets and modelled performance. Pilot data will be published as it is generated.

← Back to News & Insights