/// <summary>
/// PruneFingerprint
/// </summary>
/// <param name="qfp"></param>
/// <param name="dbfp"></param>
/// <param name="threshold"></param>
public void PruneFingerprint(
FingerprintMx qfp,
FingerprintMx dbfp,
double threshold)
{
int[] qSetBits = qfp.OnBits;
int qn = qfp.Cardinality; // # Number of bits in query fingerprint
int qmin = (int)(Math.Ceiling(qn * threshold)); // # Minimum number of bits in results fingerprints
int qmax = (int)(qn / threshold); // # Maximum number of bits in results fingerprints
int ncommon = qn - qmin + 1; // # Number of fingerprint bits in which at least one must be in common
// Get list of bits where at least one must be in result fp. Use least popular bits if possible.
//if (db.mfp_counts)
//{
// reqbits = [count['_id'] for count in db.mfp_counts.find({ '_id': { '$in': qfp} }).sort('count', 1).limit(ncommon)];
//}
//else
//{
// reqbits = qfp[:ncommon];
//}
//results = [];
//for (fp in db.molecules.find({ 'mfp.bits': { '$in': reqbits}, 'mfp.count': { '$gte': qmin, '$lte': qmax} }))
// {
// intersection = len(set(qfp) & set(fp['mfp']['bits']));
// pn = fp['mfp']['count'];
// tanimoto = float(intersection) / (pn + qn - intersection);
// if (tanimoto >= threshold)
// {
// results.append((tanimoto, fp['chembl_id'], fp['smiles']));
// }
//}
//return results;
}
/// <summary>
/// CalculateJaccardSimilarity with another FingerprintMx
/// </summary>
/// <param name="fp2"></param>
/// <returns></returns>
public double CalculateJaccardSimilarity(FingerprintMx fp2)
{
bool[] x = ToBoolArray();
bool[] y = fp2.ToBoolArray();
double sim = CalculateJaccardSimilarity(x, y);
return(sim);
}
// # Get list of bits where at least one must be in result fp. Use least popular bits if possible.
// if db.mfp_counts:
// reqbits = [count['_id'] for count in db.mfp_counts.find({'_id': {'$in': qfp
// }
//}).sort('count', 1).limit(ncommon)]
// else:
// reqbits = qfp[:ncommon]
// results = []
// for fp in db.molecules.find({'mfp.bits': {'$in': reqbits}, 'mfp.count': {'$gte': qmin, '$lte': qmax}}):
// intersection = len(set(qfp) & set(fp['mfp']['bits']))
// pn = fp['mfp']['count']
// tanimoto = float(intersection) / (pn + qn - intersection)
// if tanimoto >= threshold:
// results.append((tanimoto, fp['chembl_id'], fp['smiles']))
// return results
/// <summary>
/// Test to calculate similarity between pairs of structures (Alt syntax: Call SimSearchMx.Test [cid1] [cid2]
/// Tautomers:
/// Isotopes:
/// Neg Counterion, (Cl-) with 2 quatternary N+ with H attached in main frag:
/// Neg Counterion, (I-) with quatternary N+ (no attached H) in main frag:
/// Pos Counterion (Li+) with O- in main frag:
/// Benzene, cyclohexane:
/// StereoIsomers:
/// StereoIsomers:
/// </summary>
/// <param name="args"></param>
/// <returns></returns>
public static string Test(string args)
{
string[] sa = args.Split(' ');
string cid1 = CompoundId.Normalize(sa[0]);
string cid2 = CompoundId.Normalize(sa[1]);
int smiLen = 40;
MoleculeMx s1 = MoleculeMx.SelectMoleculeForCid(cid1);
IAtomContainer mol = CdkMol.MolfileToAtomContainer(s1.GetMolfileString());
UniChemData ucd1 = UniChemUtil.BuildUniChemData(mol);
List <FingerprintMx> fps1 = UniChemDataToFingerprintMxList(ucd1);
MoleculeMx s2 = MoleculeMx.SelectMoleculeForCid(cid2);
IAtomContainer mol2 = CdkMol.MolfileToAtomContainer(s2.GetMolfileString());
UniChemData ucd2 = UniChemUtil.BuildUniChemData(mol2);
List <FingerprintMx> fps2 = UniChemDataToFingerprintMxList(ucd2);
string fps2Smiles = (sa[0] + " / " + sa[1]).PadRight(smiLen);
string scores = "";
for (int i1 = 0; i1 < fps1.Count; i1++)
{
FingerprintMx fp1 = fps1[i1];
for (int i2 = 0; i2 < fps2.Count; i2++)
{
FingerprintMx fp2 = fps2[i2];
if (i1 == 0) // build smiles headers of cid2 frags if first cid1 frag
{
fps2Smiles += "\t" + fp2.CanonSmiles.PadRight(smiLen);
}
if (i2 == 0)
{
scores += "\r\n" + fp1.CanonSmiles.PadRight(smiLen); // include smiles at start of each line
}
float simScore = CalculateFingerprintPairSimilarityScore(fp1, fp2);
scores += "\t" + string.Format("{0:0.00}", simScore).PadRight(smiLen);
}
}
scores = fps2Smiles + scores;
FileUtil.WriteAndOpenTextDocument("SimilarityScores", scores);
return("");
}
/// <summary>
/// Calculate similarity score between a pair of FingerprintMxs
/// </summary>
/// <param name="fp1"></param>
/// <param name="fp2"></param>
/// <returns></returns>
public static float CalculateFingerprintPairSimilarityScore(
FingerprintMx fp1,
FingerprintMx fp2)
{
long[] fp1Array = fp1.ToLongArray();
OpenBitSet fp1BitSet = new OpenBitSet(fp1Array, fp1Array.Length);
int fp1Card = (int)fp1BitSet.Cardinality();
long[] fp2Array = fp2.ToLongArray();
OpenBitSet fp2BitSet = new OpenBitSet(fp2Array, fp2Array.Length);
int fp2Card = (int)fp2BitSet.Cardinality();
fp2BitSet.Intersect(fp1BitSet);
int commonCnt = (int)fp2BitSet.Cardinality();
float simScore = commonCnt / (float)(fp1Card + fp2Card - commonCnt);
return(simScore);
}
/// <summary>
/// UniChemDataToFingerprintMxList
/// </summary>
/// <param name="ucd"></param>
/// <returns>FingerprintMx list </returns>
public static List <FingerprintMx> UniChemDataToFingerprintMxList(UniChemData ucd)
{
List <FingerprintMx> fps = new List <FingerprintMx>();
FingerprintMx fp = new FingerprintMx();
fp.CdkFp = ucd.Fingerprint;
fp.CanonSmiles = ucd.CanonSmiles;
fps.Add(fp);
foreach (UniChemFIKHBHierarchy child in ucd.Children)
{
fp = new FingerprintMx();
fp.CdkFp = child.Fingerprint;
fp.CanonSmiles = child.CanonSmiles;
fps.Add(fp);
}
return(fps);
}