CHED server for predicting
soft metal binding sites in proteins


    This site uses the "CHED" algorithm to predict 3D intra-chain protein binding sites for transition metals (Zn, Fe, Mn, Cu, Ni, Co, and Ca and Mg sites that can be replaced by a transition metal).

    The algorithm searches for a triad of amino acids composed of 4 residue types (Cys, His, Glu, Asp) having ligand atoms within specific distances. It allows one target residue to rotate in rotamer space, taking into account structural rearrangements that may occur upon metal binding. A binding site is considered to be true if one or more correct amino acid ligands have been predicted.

    Machine learning algorithms are used to filter out false positives: MILD FILTER - based on the frequency of hits, yields high sensitivity (depending on metal type, 70-100% of apo sites are captured, with predictions statistically accurate in 50-70% of cases). STRINGENT FILTER - using decision tree and support vector machine technology, yields high selectivity (depending on metal type, 45-85% of apo sites are captured with predictions statistically accurate in 85-100% of cases).

Insert PDB ID code 1    and CHAIN ID           or
Your local file 2 
Click here to start    




1 If you do not know the 4-letter PDB ID code, use searching features of the OCA Browser   
2 Your local data file should follow the PDB format:
  1. The first line should have word "HEADER" at the beginning and 4-capital letter word at columns 63-66 (instead of PDB id).
  2. The lines containing protein atom coordinates must begin with the word ATOM and should have exactly the same format as in PDB.
  3. The bottom line must contain only the word END in the first three letter positions.
Reference: Babor M, Gerzon S, Rahev B, Sobolev V, Edelman M. (2008) Prediction of transition metal-binding sites from apo protein structures. Proteins 70, 208-217.
This site is part of the SPACE suite of tools for structural bioinformatics
Web site programmed and designed by Sergey Gerzon
Please mail questions/suggestions to Vladimir Sobolev