Navigating
Chemical Space
at the Speed
of Discovery
Clique Therapeutics is reimagining how drug discovery is done. Our proprietary graph network platform intelligently navigates chemical space to accelerate every stage of the discovery pipeline, delivering better candidates with fewer resources and in less time.
Drug Discovery's Most Expensive Bottleneck
Hit optimization sits at the heart of preclinical drug development, and it is where pipelines stall, budgets balloon, and timelines collapse. Despite decades of advances in high-throughput screening and computational tools, this stage remains stubbornly slow.
Existing AI approaches make it worse, not better. Drug discovery data is sparse, severely imbalanced, and riddled with noise. These are conditions that cause conventional machine learning models to fail precisely when they are needed most.
A Chemically Aware Engine
for Real-World Drug Discovery
Clique Therapeutics transforms drug discovery by treating chemistry as a relational problem, not a data volume problem. Our platform maps molecular relationships structurally using graph topology, reasoning about chemical space the way expert medicinal chemists think, but at machine speed and without the fragility of conventional AI.
Graph Topology Exploration
Our proprietary algorithms model chemical space as a relational graph, mapping structural similarity, synthetic accessibility, and biological activity in concert. This enables the engine to intelligently prioritize which regions of chemical space to explore next, compressing iteration cycles across the entire discovery pipeline.
Target-Agnostic Architecture
No crystal structure required. Our engine is broadly applicable across GPCRs, kinases, hydrolases, ion channels, transferases, oxidoreductases, and beyond, making it a platform that works across entire therapeutic portfolios, not just individual programs.
Rapid Convergence
Where conventional Active Learning requires 10 or more screening rounds and traditional medicinal chemistry demands 15 or more, Clique Tx reaches high-quality optimized leads in a fraction of the iterations. Fewer experiments, less spend, and faster timelines to the clinic.
Robust Under Real Conditions
Built for the data reality of drug discovery, not a clean benchmark. Our platform maintains full performance under extreme class imbalance, sparse datasets, and noisy experimental conditions that cause conventional AI methods to collapse.
Deep Roots Across the Drug Discovery Spectrum
Our team brings hands-on expertise spanning the full arc of drug discovery, from early target identification through preclinical candidate selection. We understand the biology, the chemistry, and the operational realities that shape successful programs.
Results That Hold Up
in Retrospective Experiments
In a rigorous retrospective benchmark against the DRD4 receptor, one of the most challenging targets in the ChEMBL library with a 1:170 active-to-decoy ratio across approximately 500,000 compounds, Clique Tx achieved a 100,000-fold improvement in potency in just two iterations.
State-of-the-art AI and Active Learning methods, given the same starting hit and the same compound library, failed to identify a single optimized lead after 10 rounds.
Across 44 retrospective experiments spanning the full breadth of druggable target classes, Clique Tx demonstrated superior performance in every category tested.
Let's Build Something
Worth Discovering
We collaborate with organizations at every stage of drug discovery, from early target selection to late-stage optimization. If you are working on a challenging program and want to move faster without compromising quality, we want to hear from you.