tech.molecules.deep.MolEncoder Maven / Gradle / Ivy

Go to download
package tech.molecules.deep;

import com.actelion.research.chem.Molecule;
import com.actelion.research.chem.StereoMolecule;
import tech.molecules.leet.chem.ChemUtils;

import java.util.*;


/**
 * Format: Each atom is encoded with the same information:
 *
 * 1. One-Hot Atom
 *
 * Information about bonds to all n other atoms. For pair of atom i and j:
 * if i and j are the same, or if i and j are not connected: all zeros
 * otherwise:
 *
 * 1.
 *
 *
 *
 *
 */
public class MolEncoder {

    private StereoMolecule m;
    int N = 32;
    int[] mappedAtomicNo = new int[] {1,6,7,8,9,15,16,17,35};

    Map encoding_AtomOneHot = new HashMap<>();

    Map encoding_BondTypeSimple = new HashMap<>();

    public MolEncoder() {
        init();
    }

    private void init() {
        for(int zi=0;zi();
        //encoding_BondTypeSimple.put(Molecule.cBondTypeSingle,oneHot(0,7));
        //encoding_BondTypeSimple.put(Molecule.cBondTypeDouble,oneHot(1,7));
        //encoding_BondTypeSimple.put(Molecule.cBondTypeTriple,oneHot(2,7));
        //encoding_BondTypeSimple.put(Molecule.cBondTypeDelocalized,oneHot(3,7));
        //encoding_BondTypeSimple.put(Molecule.cBondTypeUp,oneHot(4,7));
        //encoding_BondTypeSimple.put(Molecule.cBondTypeDown,oneHot(5,7));
        //encoding_BondTypeSimple.put(Molecule.cBondType,oneHot(6,7));
        //encoding_bondTypeSimple.put(Molecule.cBond;

    }
    public void setMolecule(StereoMolecule m) {
        this.m = m;
    }


    public List encodeMolecule() {
        m.ensureHelperArrays(Molecule.cHelperCIP);
        int[] permutation = new int[N];
        for(int zi=0;zi encodeMolecule(int[] permutation) {
        List bs = new ArrayList<>();
        for(int zi=0;zi= m.getAtoms()) {
            return new BitSet(getAtomEncodingLength() + N * getConnectionEncodingLengthPerConnection());
        }

        BitSet b_all = new BitSet( getAtomEncodingLength() + N * getConnectionEncodingLengthPerConnection() );
        BitSet b_i = encodeAtom( permutation[zi] );
        b_all.or( b_i );

        for(int zj=0;zj bsi.set(xi));
        return bsi;
    }

    public static void print(List bs, int lengthPerBitSet) {
        for(BitSet bsi : bs) {
            System.out.print("\n");
            for(int zi=0;zi encoded = ei.encodeMolecule();
        print(encoded,ei.getAtomEncodingLength()+ei.N*ei.getConnectionEncodingLengthPerConnection());
        System.out.println("");
        //for(BitSet bi : encoded) {
        //    System.out.println(bi.toString());
        //}
    }


}