Disorder quantified

We apply quantum machine learning to molecular systems that classical computation can’t model, from protein dynamics to advanced materials.

Partner with us How it works

BCR, modelled in motion. The grey backbone is surrounded by the conformations it moves through.

Chemistry · Proteins

Predicting allosteric binding sites

Allosteric sites sit away from a protein’s active pocket. They are hard to find in a single static structure, because they show up in how the protein moves. Percene models that motion and builds a residue-level map of the protein, which we use to predict where allosteric binders could act.

Allosteric binding sites Predict sites away from the active pocket where a binder could change how the protein behaves.
Residue-level resolution Work through the protein one residue pair at a time, up to the whole structure.
Modelled in motion Start from the conformations a protein moves through, not one static structure.
BCR-ABL1 Currently applied to BCR-ABL1, the fusion kinase behind chronic myeloid leukaemia.

The BCR-ABL1 residue map, zooming out from individual residue pairs to the whole protein. This is the map we use to predict allosteric binding sites.

Chemistry · Materials

A new paradigm of molecular modelling

Density functional theory carries systematic errors that no amount of compute removes. Percene models electron densities, energetics and configurations directly, lifting the accuracy ceiling on prediction of advanced materials. That opens up discovery in MOFs, polymers and larger chained chemicals — systems where the cost of being wrong is a wasted synthesis cycle.

Electron densities and energetics Model configurations and energetics directly, rather than inheriting DFT’s systematic errors.
Built for large systems MOFs, polymers and extended chained chemicals, where classical methods degrade fastest.
Discovery, not just evaluation Search configuration space for candidate structures, not only score the ones you already have.
PFAS applications Currently applied to PFAS configuration optimisation and novel candidate search.

Finalist — RSC Emerging Technologies 2026

A frontier natural orbital of [CuCO₂]⁻ under thermal motion — the strongly correlated metal-site chemistry where classical approximations hit their accuracy ceiling.

Also exploring · Data centres

The same methods, applied to GPU thermal risk.

We forecast data-centre hotspots six hours out, while there’s still time to move work away from them. We take this on case by case with operators.

Data centres →

Partner with us

We work with chemistry and materials teams, research partners and investors. If the problems on this page are yours, we’d like to hear from you.