Technology
BALEEN: Binding-Aware Ligand Embedding Encoder Network
BALEEN is our proprietary AI model for ligand-based drug discovery. Unlike traditional methods that overlook the target context, BALEEN learns binding-relevant features directly from 2D molecular inputs like SMILES. Trained on diverse ligand binding modes, it captures how molecules interact with receptors — without requiring protein data at inference. While classical fingerprints rely on predefined molecular substructures, BALEEN learns task-specific features from data, enabling a more nuanced and target-aware understanding of binding behavior.
Coming soon. Available by request.
Covalent Modeling
SieveStack's proprietary covalent modeling tool, designed to overcome challenges in drug discovery by accurately predicting covalent binding poses, is now accessible on Superbio.ai. This powerful tool integrates physics-based simulation and flexible side-chain modeling, demonstrated by its precise reproduction of experimental co-crystal structures.
Read about it here.