Help
Overview
ProtChemSI, The database of protein-chemical structural interactions includes all existing 3D structures of complexes of proteins with low molecular weight ligands from the Protein Data Bank. Ligands are defined as something present in the PubChem database. Since a protein or a chemical may present in multiple complexes with different interaction partners, ProtChemSI is essentially a network of interactions.
Biological networks are powerful tools for predicting undocumented relationships between molecules. The underlying principle is that existing interactions between molecules can be used to predict new interactions. For pairs of proteins sharing a common ligand, we use protein and chemical superimpositions combined with fast structural compatibility screens to predict whether additional compounds bound by one protein would bind the other.
Two complexes sharing a common protein of ligand can be superimposed on top of each other, i.e. a movement in the 3D space can be found that brings the atoms of the common partner from the first complex on top of the atoms of the second complex minimizing the root mean square deviation of the coordinates of the corresponding atoms. To do superimpositions, we use STAMP, a program developed by Rob Russell and Geoffrey Barton. We use this technique to propose possible new complexes.
Since protein and chemical annotations, as well as 2D structures of chemicals, are obtained from UniProt and PubChem, some error existing in those databases may propagate in ProtChemSI. 2D structures of chemicals were created from SDF files from PubChem using OpenBabel conversion tool.
Search database
On the main page, various options are available to search the database
- Search by protein name, identifier or description. In this case, all proteins whose name, identifier or description at least partially match the search term, will be displayed.
- Search by protein sequence. All proteins that are at least 30% identical to the query sequence will be displayed.
- Search by chemical name, identifier or description. Analogous to protein search.
- Search by chemical SMILES string. All chemicals that have the Tanimoto similarity score above the provided cutoff (0.9 by default) will be displayed. Additionally, all chemicals, to which the query SMILES string is an exact substructure, will be displayed.
By clicking on ‘Interactions’ on the search output page, the user queries the database for all putative interacting molecules for the search hit. For proteins, these include:
- All chemicals that the query protein contacts in known 3D structures;
- All chemicals that have the Tanimoto score of at least 0.9 to those in the direct contact;
- All chemicals that share at least one interacting protein with those in the direct contact.
For chemicals, the list includes:
- All proteins that the query chemical contacts in known 3D structures;
- All proteins that are at least 30% identical to those in the direct contact;
- All proteins that share at least one bound ligand with those in the direct contact.
The third options in both cases will be restricted to cases when a structure can be modeled based on superimposition.
Find shortest path
Since the protein-chemical 3D contacts form essentially a network of interactions, one can ask whether any two given molecules can be connected by a path in this network, and whether the molecules along this path can be superimposed to form a promising model complex. When hitting the ‘Find shortest path’ button, the user is presented with the possibility to look up the second molecule in the network and then search for the shortest path between the two molecules, evaluate the structures that can be built along it. At the moment, only the shortest path is being considered, but we plan to add sub-optimal paths.
Reference
When referring to this database, please cite:
Kalinina OV, Wichmann O, Apic G, Russell RB (2011). Combinations of protein-chemical complex structures reveal new targets for established drugs. PLoS Comp Biol 7(5): e1002043 [ PDF ] [ PubMed ] [ Full text ]
Contact
ProtChemSI was developed by Olga Kalinina and Rob Russell at the Protein Evolution group, CellNetworks, University of Heidelberg.
Please address questions, comments and bug reports to Olga Kalinina (olga.kalinina@bioquant.uni-heidelberg.de)