The ProtSynq Platform

The ProtSynq Computational Platform

Not Rosetta — you don't need a terminal. Not AlphaFold alone — structure prediction is one step. Physics-informed + learned scoring functions. Point-mutation ΔΔG prediction. Delivered as a focused research tool, not a general-purpose molecular simulation suite.

Request Access View Pricing

Three-stage scoring pipeline

From structure input to ranked variant output — each stage auditable.

Structure Preparation PDB parsing & chain assignment → residue normalization → mutable positions Scoring Ensemble Physics energy terms + learned ΔΔG models → vdW, electrostatics → transformer ΔΔG Variant Ranking Pareto stability/affinity frontier + confidence → top-5 variants → CSV + PDB diffs
Accepted Inputs

Accepted inputs

  • PDB files (.pdb) — experimentally determined structures
  • mmCIF format (.cif)
  • AlphaFold model files (.pdb from AF2/AF3 outputs) — critical for targets with no experimental structure
  • Chain-masked PDB for multi-chain complexes
  • Batch upload via API (JSON payload with base64-encoded PDB)

AlphaFold inputs are first-class citizens — most enzymes and antibodies in active engineering campaigns have AF2 models available even when crystallography hasn't been done.

Scoring Engine

How ProtSynq scores mutations

Two complementary signal streams. Physics-based terms: change in van der Waals packing, electrostatics, solvation, and backbone torsion strain for each substitution. Learned terms: a fine-tuned transformer model trained on experimentally measured ΔΔG values from mutagenesis databases.

Final ranking combines both streams under a calibrated ensemble — no single model dominates. When the two signal types disagree, a confidence interval is widened rather than suppressed. You see the disagreement.

Output Formats

Output formats

You get a ranked table, PDB diffs for each variant, and a per-residue heatmap you can pull into your LIMS — not an abstract probability score that requires interpretation.

  • Interactive ranked variant table (web portal)
  • CSV export (variant ID, mutation, ΔΔG stability, ΔΔG binding, confidence)
  • PDB diff files for each recommended variant
  • Per-residue heatmap visualization (SVG)
Integrations

Designed to fit your existing pipeline

  • Benchling integration — protein sequence versioning and project tracking
  • AlphaFold 2 / ColabFold upstream — generate a structure, feed it directly to ProtSynq
  • Standard PDB repository import — paste a PDB ID and ProtSynq fetches the structure
  • CSV export compatible with Dotmatics, Spotfire, and most LIMS — no custom adapter required

These are workflow integrations, not official partnerships. ProtSynq operates on standard file formats (PDB, CSV, SVG) — interoperability is by design.

Try it on your next protein target.

The Explore tier is free — upload your structure, run a full scan, get the top-5 variant output. No commitment required.

Request Access View Pricing