Atom Protein Structure Prediction

AI tool for the future precision medicine and precision health.

Protein Structure Prediction

Sequence/Template are constructed based on pair-wised residues attention and pair side chain relaxing. Molecular Triplet Attention model will learn the relation of protein torsion and protein distance.

We created Deep MDS Model, which uses multi-parameters to modify Linear transform results. The equivalent Graph Network Network (GNN) model refines the coordinate of atom by one to all nodes and edges attentions. Relaxing side chain is generated by Graphen protein-protein Force Filed tool. Our model uses Ca coordinate and 4D bond rotation as cost functions. It not only can predict monomer, but also can predict polymer or transmembrane protein.

In: Fasta / Out: PDB structure file


Speak to an expert today.