gistfile1.java

package biojava;

import java.util.ArrayList;
import java.util.List;

import javax.vecmath.Point3d;

import org.biojava.bio.structure.Atom;
import org.biojava.bio.structure.Chain;
import org.biojava.bio.structure.Group;
import org.biojava.bio.structure.Structure;
import org.biojava.bio.structure.io.PDBFileReader;
import org.openscience.cdk.config.AtomTypeFactory;
import org.openscience.cdk.exception.CDKException;
import org.openscience.cdk.graph.rebond.RebondTool;
import org.openscience.cdk.interfaces.IAtom;
import org.openscience.cdk.interfaces.IChemObjectBuilder;
import org.openscience.cdk.interfaces.IMolecule;
import org.openscience.cdk.nonotify.NoNotificationChemObjectBuilder;
import org.openscience.cdk.smiles.SmilesGenerator;


public class LigandExtractor {
    
    private static IChemObjectBuilder builder = 
        NoNotificationChemObjectBuilder.getInstance();
    
    public static List<IMolecule> extractLigands(String path) throws CDKException {
        List<IMolecule> ligands = new ArrayList<IMolecule>();
       
        RebondTool tool = new RebondTool(2.0, 0.5, 0.5);
        try {
            String code = new String(path).replace(".pdb", "");

            PDBFileReader pdbreader = new PDBFileReader();
            pdbreader.setPath(".");
            pdbreader.setParseSecStruc(true);// parse the secondary structure
                                             // information from PDB file
            pdbreader.setAlignSeqRes(true); // align SEQRES and ATOM records
            pdbreader.setAutoFetch(false); // fetch PDB files from web if they
                                          // can't be found locally

            Structure struc = pdbreader.getStructureById(code);

            int modelnr = 0; // also is 0 if structure is an XRAY structure.

            List<Chain> chains = struc.getChains(modelnr);
            
            for (Chain cha : chains) {
                List<Group> hgr = cha.getAtomGroups("hetatm");
                LigandExtractor.makeLigandsFromGroups(hgr, ligands);
            }

        } catch (Exception e) {
            e.printStackTrace();
        }
        AtomTypeFactory pdbFactory = AtomTypeFactory.getInstance(
                "org/openscience/cdk/config/data/jmol_atomtypes.txt", builder);
        for (IMolecule ligand : ligands) {
            try {
                for (IAtom atom : ligand.atoms()) {
                    pdbFactory.configure(atom);
                }
                tool.rebond(ligand);
            } catch (Exception e) {
                e.printStackTrace();
            }
        }
        
        return ligands;
    }
    
    public static void makeLigandsFromGroups(
            List<Group> groups, List<IMolecule> ligands) {
        
        for (Group group : groups) {
            IMolecule molecule = builder.newMolecule();
            for (Atom atom : group.getAtoms()) {
                String elementSymbol = 
                    LigandExtractor.convertNameToElementSymbol(atom.getName());
                
                Point3d p = new Point3d(atom.getX(), atom.getY(), atom.getZ());
                molecule.addAtom(builder.newAtom(elementSymbol, p));
            }
            ligands.add(molecule);
        }
    }
    
    public static String convertNameToElementSymbol(String atomName) {
       return atomName.substring(0, 1);
    }
    
    public static void main(String[] args) {
        try {
            SmilesGenerator generator = new SmilesGenerator();
            for (IMolecule ligand : new LigandExtractor().extractLigands("1hxb.pdb")) {
                int i = ligand.getAtomCount();
                if (i > 1) {
                    System.err.println(generator.createSMILES(ligand));
                }
            }
        } catch (CDKException c) {
            c.printStackTrace();
        }
    }
}