Skip to content
Home/Quantum Computing News/SandboxAQ Ranks Drug Candidates for $1 per 1,000
IndustryUse CasesChemistry

SandboxAQ Ranks Drug Candidates for $1 per 1,000

SandboxAQ's AQPotency tool predicts drug efficacy without a solved protein structure and reports a confidence interval with each score.

FreeQuantumComputing
·· 4 min read

SandboxAQ launched AQPotency, a ranking tool for drug candidates costing as little as $1 per 1,000 comparisons. The Large Quantitative Model predicts how well a molecule will bind and act without a solved 3D structure of the disease target. Older virtual screening methods cost more and need structural data many targets lack.

The structural-data bottleneck

Many promising drug targets have no solved protein structure, and programs stall when a structural map is missing. SandboxAQ says AQPotency predicts efficacy without a solved structure, opening targets earlier tools skipped. The model reports a confidence interval with each prediction, so researchers know when a score is reliable enough to act on.

Two directions

AQPotency works in reverse too. Given a molecule with a beneficial effect and an unknown mechanism, the tool scans proteins to find likely interaction targets. Professor Dario Alessi at the University of Dundee cited the models as useful for Parkinson's work, and Columbia University researchers used the approach on SV2C membrane targets.

What the price point hides

SandboxAQ is an AI and sensing company spun out of Alphabet, and AQPotency is classical machine learning, not a quantum processor. The $1 per 1,000 price and the confidence reporting are the claims worth tracking, both unverified outside SandboxAQ's own announcement. For context on the gap between quantum machine learning promises and shipped results, see our quantum machine learning reality check.