Multivalent ion binding site identification with structure-based deep learning
Nature
Read full post
Researchers developed BiteNetI, a deep learning model using 3D convolutional neural networks to identify binding sites and residues for 14 key biological ions in proteins. Trained on over 10,000 protein-ion complexes, it outperforms existing predictors and AlphaFold3 in accuracy, especially for ions like Ca2+, Na+, and K+.
