GeoBind

Geometric deep learning for protein functional binding sites prediction

1. Input PDB file [Note 1]


2. Input Chains of PDB file [Note 2]

Chains

3. Select ligand type [Note 3]

4. E-mail for informing

E-mail

5. Run GeoBind


Notes:

Note 1. The file to upload requires the PDB format file (.pdb or .cif). The PDB file may contain multiple chains. Leave it blank and submit, the server will yield an example of prediction result.

Note 2. Prediction of multimer is supported. If monomer, fill with the single Chain ID, such as "A". If multimer, fill with Chain IDs seperated by '_', such as "A_B_C". If blank, all chains in the PDB file are counted. We recommend the number of chains do not exccede 8 for the computational limitation by msms program.

Note 3. GeoBind now can predict the seven types of ligands referring to our paper. We may extend GeoBind to other types of ligand binding prediction function as long as we collect dataset and pre-train GeoBind.

Note 4. For local deployment, the source code is available at www.github.com/zpliulab/GeoBind.
Webserver in construction (v0.1).

© 2022 Zhi-Ping Liu lab, Shandong University.