Protein Design Solutions offers the PhiPI platform, which uses all‑atom molecular dynamics simulations combined with a graph neural network to generate residue‑level hydration thermodynamics maps for protein therapeutics. The platform identifies aggregation‑prone and high‑viscosity regions, ranks mutation candidates to improve solubility and manufacturability, and delivers results in visual overlays and exportable tables for integration into CMC workflows.
Funding
Funding not disclosed
Founders
Product
Problem
Biopharmaceutical candidates frequently encounter aggregation, high viscosity, and other developability challenges that delay formulation, scale‑up, and regulatory timelines. These issues often surface late in the CMC workflow, forcing costly reformulation cycles and jeopardizing clinical timelines.
Solution
Protein Design Solutions addresses these risks with the PhiPI platform, which combines physics‑informed all‑atom protein‑water simulations and a proprietary graph neural network (GNN) to produce residue‑level hydration thermodynamics maps. The system quantifies hidden hydrophobic patches and other high‑risk surface features, then ranks mutation candidates that mitigate aggregation and viscosity without compromising activity. Results are delivered as visual patch overlays, ranked tables, and concise mutation shortlists that can be imported directly into existing biophysical assay pipelines. By delivering high‑fidelity predictions in seconds, PhiPI enables rapid design‑build‑test iterations and early de‑risking of protein therapeutics. The platform is offered via subscription or per‑project analysis, allowing teams to align costs with project scope.
Target Audience
Primary customers are biotech and pharmaceutical CMC groups, protein engineering teams, and formulation scientists who need quantitative, residue‑specific developability insights during early candidate selection.
Features
- All‑atom phi‑ensemble molecular dynamics simulations that capture protein‑water interaction energetics at the residue level
- Proprietary GNN trained on physics‑derived ground truth, delivering high‑accuracy risk maps in seconds
- Residue‑level heatmap visualizations highlighting hidden hydrophobicity and aggregation‑prone sites
- Automated ranking of single‑point mutations with predicted impact on solubility, viscosity, and manufacturability
- Exportable patch overlays and mutation tables compatible with common assay data formats (e.g., CSV, JSON) and laboratory information management systems (LIMS)
- Secure cloud‑based analytics pipeline with end‑to‑end encryption for proprietary protein sequences
- API endpoints for integration into in‑house design workflows and automated variant screening pipelines