Virtual screening for hit identification
How our system works
MSLHFLYYCSEPTLDVKIAFCQGFD…
Step 1: Input protein sequence and compound
Step 2: Predict 3D complex structure
Step 3: Assess affinity strength on 3D structure
72.3%
Step 4: Return binding likelihood
How we work with you
1 Feasibility assessment
Before a full campaign, we perform a pre-study to evaluate the target’s amenability to these specific computational methods.
2 Collaborative review
We review feasibility data with your team, proceeding to a full-scale screening only when the evidence supports a favourable outcome.
3 Large-scale virtual screening
We execute the screening against internal or third-party compound libraries. Our system models the 3D structure of each protein-ligand combination and analyses the resulting poses to assess binding probability.
4 Results delivery
We deliver a shortlist of top-ranked compounds, prioritising chemical diversity to maximise the success rate of subsequent lead optimisation.
What sets our system apart
Induced-fit co-folding
Our methods identify hits requiring side-chain or backbone flexibility – novel hit candidates that are often inaccessible to traditional docking tools that lack native support for pocket flexibility.
Support for novel targets
The system remains effective even in the absence of experimental structural data. This expands the reach of virtual screening to novel or difficult-to-crystallise targets.
Physics-informed AI
Our AI system is trained to predict binding based on learned biophysical interactions – such as hydrogen bonding, π-π interactions and hydrophobic effects. This ensures that predicted hits are not just statistically likely, but physically plausible.
Hybrid validation
We combine the latest machine learning models with high-performance physics-based tools. For instance, traditional docking and molecular modelling can be used to validate hits prior to laboratory testing.