Industry
Drug Discovery & Life Sciences
Screening and simulation work where the candidate space is larger than any lab schedule can cover.
Problems we address
- Candidate libraries far too large to screen experimentally, and selection criteria that are mostly heuristic
- Binding and interaction predictions confident enough to publish but not to commit a synthesis budget to
- Target identification that depends on data spread across incompatible sources
- Simulation runs that return after the go/no-go decision has already been made
How we approach it
Related research
Maturity labels are not decoration. Most of this is at concept stage.
Accelerated Drug Screening
Virtual screening has to be cheap enough to run across a whole library and accurate enough not to throw away good molecules. Docking scores track binding affinity loosely, so shortlists carry false positives you can measure and false negatives nobody ever finds out about. We are working on scoring for the middle of the funnel, tested by whether it ranks known binders highly without being told.
Molecular Optimization
Conformation, energy, binding and molecular-property optimisation.
Targeted Drug Design
Binding affinity, selectivity and off-target risk concepts for targeted compounds.
Targeting and Prediction
Protein/biological target identification and interaction prediction.
Quantum Powered Protein Design
Machine learning has transformed structure prediction; design — going from a desired function to a sequence — is still much harder, and the gap between a predicted structure and a working protein is routinely surprising. We are looking at conformational search and interaction scoring where classical scoring functions are known to be weak. Nothing here counts without wet-lab validation, which is a field where computational and experimental results diverge often enough that nothing else does.