Predict per-residue lDDT scores for
3D protein structures using GCPNet, our newly-developed graph neural network
for learning efficiently from 3D biomolecular structures.
View on GitHub
For compatibility purposes, this application supports returning CAMEO-style per-residue structural error predictions (approximately in Angstroms). Leaving this option disabled will instead return AlphaFold-style per-residue plDDT.