Use cases · Pharma · life sciences
Find where a protein can be drugged before the assay
Candidate allosteric sites read from a structure file alone, with the site model validated against exact references. A method result: no drug-ranking gain is claimed.
Impact
A ranked list of candidate sites from structure alone
Baseline: an assay campaign per target
Condition: input is a structure file; no assay data required
A site model you can check
Baseline: pairwise contact models, which miss collective effects
Condition: the model agrees with its exact reference within 1.4%
The problem
Allosteric sites, the places away from the active site where a drug can still switch a protein, are usually found by assay campaigns that take months and are run on every target in turn.
Our approach
We read the protein’s contact network as a coupled system and ranked residues by how strongly they carry the response, from a PDB or mmCIF structure alone. The site model was checked against exact references and against known biology.
The evidence
- Collective-response model within 1.4% of exact at the validated size (quantum hardware plus simulation).
- Area under the curve up to 0.89 against known biology.
- On the independent public benchmark we do not lead the current best method; we state that plainly.
- Not claimed: a drug-ranking gain.
Status: Quantum hardware + Simulation. Measured on quantum hardware and in simulation, validated against exact references. Method result only.