Turn a Question Into a Structured Project
Pick a project type, and Surface 2 sets up the workflow for you — inputs, models, and a report template ready to fill in as you go.
A step-by-step project flow: choose a disease area, frame a research question, run the relevant models, and build out a structured write-up in the Notebook and Report Builder as you go. Each step saves automatically so you can pick up where you left off.
Pick a gene and a set of variants; Surface 2 predicts the structural impact of each one using Alpha and flags which are most likely to affect function.
Model how a designed molecule (e.g. a CAR-T receptor) binds to its target using Gamma, and compare binding strength across design variants.
Run the same target through two different models (e.g. Beta vs Delta) and use Model Compare to evaluate which one performs better for your case.
- ▸Relevant background literature is pulled in automatically based on your topic
- ▸Every run is saved with its model version and settings, so your results are reproducible
- ▸A Notebook is created for this project so you can record hypotheses and findings as you go
Have a gene variant you're investigating? Enter the gene and the variant (e.g. a missense mutation), and Surface 2 will predict both the normal and variant structures, score how likely the variant is to disrupt protein function, and generate a short interpretation summary — including what's known about related conditions.
- Side-by-side structure comparison
- A pathogenicity-style score for the variant
- A written summary covering structural effect and related disease associations
- Suggested next steps for follow-up
Curious how a genetic variant might change someone's response to a drug? Enter a variant (e.g. in a drug-metabolizing enzyme) and a drug name, and Surface 2 will model how the drug docks against both the normal and variant versions of the protein — showing whether the variant likely changes how well the drug binds. Candidate compound data is pulled live from ChEMBL's public bioactivity database — these are real experimentally-measured values, not AI-generated predictions.
- Docking comparison between normal and variant protein
- A summary of whether binding strength, position, or accessibility changes
- Notes you can save for future reference