Chemistry and materials science
Screen thousands of compositions before you make one.
Antinuous ranks candidate materials from databases and papers and runs atomistic simulations across many structures in parallel. Predictions are labeled, and each value links to its paper or run.
Use cases
Screen candidate materials before making them
There are far more possible compositions than any lab can make.
You ask
Find lithium-containing oxides likely to be stable and to conduct lithium ions well. Rank the top 20 and show how confident each prediction is.
What it does
- Antinuous collects candidate compositions and reported properties from papers and public sources.
- It installs open-source packages such as pymatgen and a machine-learned interatomic potential in its code workspace.
- It runs stability and property estimates on each candidate and records which numbers are predictions.
- It ranks the list and links each row to its calculation or source.
You get
A ranked sheet with predicted properties, uncertainty and sources for each composition.
New crystals predicted2.2 million
Predicted to be stable380,000
Made by outside labs736
Google DeepMind's GNoME results, from predicted crystals to ones made in labs.
Who it's for
Materials scientistsScreen candidate materials before you make them.
Synthetic chemistsCompare published synthesis routes with each value linked to its paper.
Process chemistsFind safer solvent replacements with cited hazard data.
R&D leadsPlan your next experiment from past results.
Works with
PubChem
ChEMBL
RCSB PDB
arXiv
GitHub
Google Slides
PowerPoint