• drug target;
  • ligand matching;
  • potential target database;
  • virtual screen

Drugs interacting with off-target proteins would bring about side-effects. The identification of the proteins that a drug can bind is thus valuable for evaluating its side-effects. We established a system based on PDB database for screening for proteins a drug could bind. Firstly, all complexes in the PDB database were sorted by species; then, a ligand database was established by extracting ligands from the structure data files. Secondly, all proteins were clustered according to their sequence similarity with the protein originally bound with the ligand in PDB. To search the potential target proteins of a drug, the query drug structure is compared with all ligands in the database to obtain similar scores. Ligands with similar sores greater than a certain threshold were flagged. Protein clusters associating with these ligands would be considered as potential targets of the query drug. To test the reliability of this approach, three drugs from DrugBank were used to search for their binding proteins by our method. The results showed that all the corresponding target proteins were found. The method presented here was rapid, scalable and could be used for high efficient drug side-effects analysis.