public override void TestBug934819()
{
IAtomContainer subStructure = Bug934819_1();
IAtomContainer superStructure = Bug934819_2();
AtomContainerManipulator.PercieveAtomTypesAndConfigureAtoms(superStructure);
AtomContainerManipulator.PercieveAtomTypesAndConfigureAtoms(subStructure);
AddImplicitHydrogens(superStructure);
AddImplicitHydrogens(subStructure);
IFingerprinter fpr = new PubchemFingerprinter();
IBitFingerprint superBits = fpr.GetBitFingerprint(superStructure);
IBitFingerprint subBits = fpr.GetBitFingerprint(subStructure);
Assert.IsTrue(BitArrays.Equals(
AsBitSet(9, 10, 14, 18, 19, 33, 143, 146, 255, 256, 283, 284, 285, 293, 301, 332, 344, 349, 351,
353, 355, 368, 370, 371, 376, 383, 384, 395, 401, 412, 416, 421, 423, 434, 441, 446, 449, 454,
455, 464, 470, 471, 480, 489, 490, 500, 502, 507, 513, 514, 516, 520, 524, 531, 532, 545, 546,
549, 552, 556, 558, 564, 570, 586, 592, 599, 600, 607, 633, 658, 665),
subBits.AsBitSet()));
Assert.IsTrue(BitArrays.Equals(
AsBitSet(9, 10, 11, 14, 18, 19, 33, 34, 143, 146, 150, 153, 255, 256, 257, 258, 283, 284, 285, 293,
301, 332, 344, 349, 351, 353, 355, 368, 370, 371, 374, 376, 383, 384, 395, 401, 412, 416, 417,
421, 423, 427, 434, 441, 446, 449, 454, 455, 460, 464, 470, 471, 479, 480, 489, 490, 500, 502,
507, 513, 514, 516, 520, 524, 531, 532, 545, 546, 549, 552, 556, 558, 564, 570, 578, 582, 584,
586, 592, 595, 600, 603, 607, 608, 633, 634, 640, 658, 660, 664, 665, 668, 677, 678, 683),
superBits.AsBitSet()));
}
public void TestAnonFingerprints()
{
var expected = new BitArray[]
{
ParseBitSet("{148, 206, 392, 542, 637, 742, 752, 830}"),
ParseBitSet("{148, 206, 392, 542, 637, 742, 752, 830}"),
ParseBitSet("{148, 206, 392, 542, 637, 742, 752, 830}"),
ParseBitSet("{148, 206, 392, 542, 637, 742, 752, 830}"),
ParseBitSet("{148, 206, 392, 542, 637, 742, 752, 830}"),
ParseBitSet("{148, 206, 392, 542, 637, 742, 752, 830}"),
ParseBitSet("{148, 206, 392, 542, 637, 742, 752, 830}"),
ParseBitSet("{148, 206, 392, 542, 637, 742, 752, 830}"),
ParseBitSet("{148, 206, 392, 542, 637, 742, 752, 830}"),
ParseBitSet("{148, 206, 392, 542, 637, 742, 752, 830}"),
};
var filename = "NCDK.Data.MDL.fingerprints_from_modelbuilder3d.sdf";
var ins = this.GetType().Assembly.GetManifestResourceStream(filename);
var data = TemplateExtractor.MakeFingerprintsFromSdf(true, true,
new Dictionary <string, int>(), new StreamReader(ins), 10);
var obj = new QueryChemObject();
var dummy = obj.Builder;
for (int i = 0; i < data.Count; i++)
{
IBitFingerprint bs = data[i];
Assert.IsTrue(Compares.AreEqual(expected[i], bs.AsBitSet()));
}
}
public override void TestBug706786()
{
IAtomContainer superStructure = Bug706786_1();
IAtomContainer subStructure = Bug706786_2();
AddImplicitHydrogens(superStructure);
AddImplicitHydrogens(subStructure);
// SMARTS is now correct and D will include H atoms, CDK had this wrong
// for years (had it has non-H count). Whilst you can set the optional
// SMARTS flavor CDK_LEGACY this is not correct
AtomContainerManipulator.SuppressHydrogens(superStructure);
AtomContainerManipulator.SuppressHydrogens(subStructure);
IFingerprinter fpr = GetBitFingerprinter();
IBitFingerprint superBits = fpr.GetBitFingerprint(superStructure);
IBitFingerprint subBits = fpr.GetBitFingerprint(subStructure);
Assert.IsTrue(BitArrays.Equals(
AsBitSet(0, 11, 13, 17, 40, 48, 136, 273, 274, 278, 286, 294, 299, 301, 304, 306),
superBits.AsBitSet()));
Assert.IsTrue(BitArrays.Equals(
AsBitSet(1, 17, 273, 274, 278, 294, 306),
subBits.AsBitSet()));
}
public void TestFingerprint()
{
var parser = CDK.SmilesParser;
IFingerprinter printer = new EStateFingerprinter();
IBitFingerprint bs1 = printer.GetBitFingerprint(parser.ParseSmiles("C=C-C#N"));
IBitFingerprint bs2 = printer.GetBitFingerprint(parser.ParseSmiles("C=CCC(O)CC#N"));
Assert.AreEqual(79, printer.Length);
Assert.IsTrue(bs1[7]);
Assert.IsTrue(bs1[10]);
Assert.IsTrue(FingerprinterTool.IsSubset(bs2.AsBitSet(), bs1.AsBitSet()));
}
public override void TestBug706786()
{
IAtomContainer superStructure = Bug706786_1();
IAtomContainer subStructure = Bug706786_2();
AtomContainerManipulator.PercieveAtomTypesAndConfigureAtoms(superStructure);
AtomContainerManipulator.PercieveAtomTypesAndConfigureAtoms(subStructure);
AddImplicitHydrogens(superStructure);
AddImplicitHydrogens(subStructure);
IFingerprinter fpr = new MACCSFingerprinter();
IBitFingerprint superBits = fpr.GetBitFingerprint(superStructure);
IBitFingerprint subBits = fpr.GetBitFingerprint(subStructure);
Assert.IsTrue(BitArrays.Equals(
AsBitSet(53, 56, 65, 71, 73, 88, 97, 104, 111, 112, 126, 130, 136, 138, 139, 140, 142, 143,
144, 145, 148, 149, 151, 153, 156, 158, 159, 161, 162, 163, 164),
superBits.AsBitSet()));
Assert.IsTrue(BitArrays.Equals(
AsBitSet(56, 97, 104, 108, 112, 117, 131, 136, 143, 144, 146, 151, 152, 156, 161, 162, 163, 164),
subBits.AsBitSet()));
}
public override void TestBug706786()
{
IAtomContainer superStructure = Bug706786_1();
IAtomContainer subStructure = Bug706786_2();
AtomContainerManipulator.PercieveAtomTypesAndConfigureAtoms(superStructure);
AtomContainerManipulator.PercieveAtomTypesAndConfigureAtoms(subStructure);
AddImplicitHydrogens(superStructure);
AddImplicitHydrogens(subStructure);
// SMARTS is now correct and D will include H atoms, CDK had this wrong
// for years (had it has non-H count). Whilst you can set the optional
// SMARTS flavor CDK_LEGACY this is not correct
AtomContainerManipulator.SuppressHydrogens(superStructure);
AtomContainerManipulator.SuppressHydrogens(subStructure);
IFingerprinter fpr = new EStateFingerprinter();
IBitFingerprint superBits = fpr.GetBitFingerprint(superStructure);
IBitFingerprint subBits = fpr.GetBitFingerprint(subStructure);
Assert.IsTrue(BitArrays.Equals(AsBitSet(6, 11, 12, 15, 16, 18, 33, 34, 35), superBits.AsBitSet()));
Assert.IsTrue(BitArrays.Equals(AsBitSet(8, 11, 16, 35), subBits.AsBitSet()));
}
/// <summary>
/// BuildTest
/// </summary>
public static void BuildTest()
{
CircularFingerprinter cfp = null;
CircularFingerprinterClass FpClass = CircularFingerprinterClass.ECFP6; // FP diameter
int FpLen = 2048; // folded binary fp length
IAtomContainer mol, mol2;
//string molfile = FileUtil.ReadFile(@"C:\Download\CorpId-12345.mol");
//java.io.StringReader sr = new java.io.StringReader(molfile);
//if (Lex.Contains(molfile, "v2000"))
// cor = new MDLV2000Reader(sr);
//else
// cor = new MDLV3000Reader(sr);
//cor.setReaderMode(IChemObjectReader.Mode.RELAXED);
//ac = (IAtomContainer)cor.read(new AtomContainer());
//cor.close();
FpClass = CircularFingerprinterClass.ECFP4; // debug
cfp = new CircularFingerprinter(FpClass, FpLen);
StreamReader reader = new StreamReader(@"C:\Download\CorpId-12345.mol");
//FileReader FileReader = new FileReader(@"C:\Download\V3000 Mols.sdf");
EnumerableSDFReader rdr = new EnumerableSDFReader(reader.BaseStream, ChemObjectBuilder.Instance);
rdr.ReaderMode = ChemObjectReaderMode.Relaxed;
IEnumerator <IAtomContainer> cursor = rdr.GetEnumerator();
while (cursor.MoveNext())
{
mol = cursor.Current;
mol = CdkMol.GetLargestMoleculeFragment(mol);
ICountFingerprint cfp1 = cfp.GetCountFingerprint(mol); // get hash values and counts for each
cfp.Calculate(mol);
int fpCount = cfp.FPCount;
for (int fpi = 0; fpi < fpCount; fpi++) // gets
{
CircularFingerprint cfp2 = cfp.GetFP(fpi); // gets hash, iteration and lists of atoms (dups appear multiple times)
}
IBitFingerprint bfp = cfp.GetBitFingerprint(mol);
BitArray bs = bfp.AsBitSet();
int bsCard = bfp.Cardinality;
long bsSize = bfp.Length;
continue;
}
reader.Close();
return;
//java.io.StringReader sr = new java.io.StringReader(molfile);
//AtomContainer mol = new AtomContainer();
//mol.addAtom(new Atom("C"));
//mol.addAtom(new Atom("H"));
//mol.addAtom(new Atom("H"));
//mol.addAtom(new Atom("H"));
//mol.addAtom(new Atom("H"));
//mol.addBond(new Bond(mol.getAtom(0), mol.getAtom(1)));
//mol.addBond(new Bond(mol.getAtom(0), mol.getAtom(2)));
//mol.addBond(new Bond(mol.getAtom(0), mol.getAtom(3)));
//mol.addBond(new Bond(mol.getAtom(0), mol.getAtom(4)));
//FileReader FileReader = new FileReader(@"C:\Download\CorpId-12345.mol");
//MolReader mr = new MolReader(FileReader, DefaultChemObjectBuilder.getInstance());
//java.io.StringReader sr = new java.io.StringReader(molfile);
//IMol m = (IMol)mr.next();
//FileReader.close();
}