Search possibilities
A protein model scans PETase changes and builds a shortlist of leads.
The open PETase hunt
PETase can break down PET plastic. We’re searching for better versions—and your computer can help scan the possibilities.
This is a computational hunt: AI helps choose candidates to investigate. Only lab tests can prove an enzyme works better.
One search, many computers
Start a scan here. Opt in to let trusted computers help. Every candidate and score stays open to inspect.
The candidate finder
Start with 73 distinct protein examples from PETML. Its model learned from published PETase activity measurements and ranks each candidate against the original IsPETase.
The model runs here in your browser. No account needed.
This is a first search pool supplied by the model’s authors. Its ranking is a research lead, not a measured result or a new discovery. See the model and source data ↗
Open shortlist
Run the activity model to reveal the strongest leads in this example pool.
Higher scores mean the literature-trained model favors that sequence over IsPETase. They are not a measured activity or a guaranteed chance of improvement.
A search anyone can join
No lab coat required. Just a browser and a little spare compute.
A protein model scans PETase changes and builds a shortlist of leads.
Choose to share spare compute. Square65 splits the search into small jobs.
Share the most interesting lead. The community can inspect the same evidence.
Real sequence · real model
Scan 152 possible PETase changes. The model ranks which ones fit the protein sequence best; that gives us leads to investigate, not proof of better plastic breakdown.
First run downloads the pinned model to your browser. Then opt in to share the work across your computers. Pooled scores are checked against your local run. No lab activity or speedup is implied.
Ready when you are. Start the search on this computer.
Positive values mean the model favors that residue over the reference at the same position. They do not predict a fold change in PET breakdown.