use of de.ipbhalle.metfraglib.FastBitArray in project MetFragRelaunched by ipb-halle.
the class CalculateHydrogenDeuteriumScore method getEqualPeakPairs.
/**
* @param massesHydrogen
* @param massesDeuterium
* @param ppm
* @param abs
* @return
*/
public static int[][] getEqualPeakPairs(double[] massesHydrogen, double[] massesDeuterium, double ppm, double abs) {
ArrayList<Integer> equalPeakPairHydrogenIndeces = new ArrayList<Integer>();
ArrayList<Integer> equalPeakPairDeuteriumIndeces = new ArrayList<Integer>();
FastBitArray hydrogensAnnotated = new FastBitArray(massesHydrogen.length);
FastBitArray deuteriumsAnnotated = new FastBitArray(massesDeuterium.length);
double[] absDeviationHydrogens = new double[massesHydrogen.length];
double[] absDeviationDeuterium = new double[massesDeuterium.length];
for (int i = 0; i < massesHydrogen.length; i++) {
double mzabs = MathTools.calculateAbsoluteDeviation(massesHydrogen[i], ppm) + abs;
for (int j = 0; j < massesDeuterium.length; j++) {
double deviation = Math.abs(massesHydrogen[i] - massesDeuterium[j]);
if (deviation <= mzabs) {
if (hydrogensAnnotated.get(i) && deuteriumsAnnotated.get(j))
continue;
else if (hydrogensAnnotated.get(i)) {
if (deviation < absDeviationHydrogens[i]) {
int index = equalPeakPairHydrogenIndeces.indexOf(i);
equalPeakPairHydrogenIndeces.remove(index);
equalPeakPairDeuteriumIndeces.remove(index);
equalPeakPairHydrogenIndeces.add(i);
equalPeakPairDeuteriumIndeces.add(j);
absDeviationHydrogens[i] = deviation;
absDeviationDeuterium[j] = deviation;
}
} else if (deuteriumsAnnotated.get(j)) {
if (deviation < absDeviationDeuterium[j]) {
int index = equalPeakPairDeuteriumIndeces.indexOf(j);
equalPeakPairHydrogenIndeces.remove(index);
equalPeakPairDeuteriumIndeces.remove(index);
equalPeakPairHydrogenIndeces.add(i);
equalPeakPairDeuteriumIndeces.add(j);
absDeviationHydrogens[i] = deviation;
absDeviationDeuterium[j] = deviation;
}
} else {
equalPeakPairHydrogenIndeces.add(i);
equalPeakPairDeuteriumIndeces.add(j);
absDeviationHydrogens[i] = deviation;
absDeviationDeuterium[j] = deviation;
}
hydrogensAnnotated.set(i);
deuteriumsAnnotated.set(j);
}
}
}
int[][] pairedIndeces = new int[equalPeakPairHydrogenIndeces.size()][2];
for (int i = 0; i < equalPeakPairHydrogenIndeces.size(); i++) {
pairedIndeces[i][0] = equalPeakPairHydrogenIndeces.get(i);
pairedIndeces[i][1] = equalPeakPairDeuteriumIndeces.get(i);
}
return pairedIndeces;
}
use of de.ipbhalle.metfraglib.FastBitArray in project MetFragRelaunched by ipb-halle.
the class InChIDeuteriumGeneration method getAromaticAtoms.
/**
* @param molecule
* @return
*/
public static FastBitArray getAromaticAtoms(IAtomContainer molecule) {
Aromaticity arom = new Aromaticity(ElectronDonation.cdk(), Cycles.cdkAromaticSet());
FastBitArray aromaticAtoms = new FastBitArray(molecule.getAtomCount());
try {
AtomContainerManipulator.percieveAtomTypesAndConfigureAtoms(molecule);
arom.apply(molecule);
Set<IBond> aromaticBonds = arom.findBonds(molecule);
Iterator<IBond> it = aromaticBonds.iterator();
while (it.hasNext()) {
IBond bond = it.next();
for (int k = 0; k < bond.getAtomCount(); k++) aromaticAtoms.set(molecule.indexOf(bond.getAtom(k)));
}
} catch (CDKException e) {
e.printStackTrace();
}
return aromaticAtoms;
}
use of de.ipbhalle.metfraglib.FastBitArray in project MetFragRelaunched by ipb-halle.
the class SDFDeuteriumGeneration method getAromaticAtoms.
/**
* @param molecule
* @return
*/
public static FastBitArray getAromaticAtoms(IAtomContainer molecule) {
Aromaticity arom = new Aromaticity(ElectronDonation.cdk(), Cycles.cdkAromaticSet());
FastBitArray aromaticAtoms = new FastBitArray(molecule.getAtomCount());
try {
AtomContainerManipulator.percieveAtomTypesAndConfigureAtoms(molecule);
arom.apply(molecule);
Set<IBond> aromaticBonds = arom.findBonds(molecule);
Iterator<IBond> it = aromaticBonds.iterator();
while (it.hasNext()) {
IBond bond = it.next();
for (int k = 0; k < bond.getAtomCount(); k++) aromaticAtoms.set(molecule.indexOf(bond.getAtom(k)));
}
} catch (CDKException e) {
e.printStackTrace();
}
return aromaticAtoms;
}
use of de.ipbhalle.metfraglib.FastBitArray in project MetFragRelaunched by ipb-halle.
the class SmilesDeuteriumGeneration method getAromaticAtoms.
/**
* @param molecule
* @return
*/
public static FastBitArray getAromaticAtoms(IAtomContainer molecule) {
Aromaticity arom = new Aromaticity(ElectronDonation.cdk(), Cycles.cdkAromaticSet());
FastBitArray aromaticAtoms = new FastBitArray(molecule.getAtomCount());
try {
AtomContainerManipulator.percieveAtomTypesAndConfigureAtoms(molecule);
arom.apply(molecule);
Set<IBond> aromaticBonds = arom.findBonds(molecule);
Iterator<IBond> it = aromaticBonds.iterator();
while (it.hasNext()) {
IBond bond = it.next();
for (int k = 0; k < bond.getAtomCount(); k++) aromaticAtoms.set(molecule.indexOf(bond.getAtom(k)));
}
} catch (CDKException e) {
e.printStackTrace();
}
return aromaticAtoms;
}
use of de.ipbhalle.metfraglib.FastBitArray in project MetFragRelaunched by ipb-halle.
the class AutomatedLossFingerprintAnnotationScoreInitialiser method postProcessScoreParameters.
public void postProcessScoreParameters(Settings settings) throws AtomTypeNotKnownFromInputListException, Exception {
CombinedSingleCandidateMetFragProcess[] processes = (CombinedSingleCandidateMetFragProcess[]) settings.get(VariableNames.METFRAG_PROCESSES_NAME);
// fingerprints not seen in training
MassToFingerprintsHashMap lossMassToFingerprints = new MassToFingerprintsHashMap();
MassToFingerprintGroupListCollection lossToFingerprintGroupListCollection = (MassToFingerprintGroupListCollection) settings.get(VariableNames.LOSS_TO_FINGERPRINT_GROUP_LIST_COLLECTION_NAME);
Double mzppm = (Double) settings.get(VariableNames.RELATIVE_MASS_DEVIATION_NAME);
Double mzabs = (Double) settings.get(VariableNames.ABSOLUTE_MASS_DEVIATION_NAME);
int ionmode = (Integer) settings.get(VariableNames.PRECURSOR_ION_MODE_NAME);
boolean ispositive = (Boolean) settings.get(VariableNames.IS_POSITIVE_ION_MODE_NAME);
double adductMass = Constants.getIonisationTypeMassCorrection(Constants.ADDUCT_NOMINAL_MASSES.indexOf(ionmode), ispositive);
double precursorMass = (Double) settings.get(VariableNames.PRECURSOR_NEUTRAL_MASS_NAME);
double ionmass = MathTools.round(precursorMass + adductMass);
Fingerprint fingerprint = new Fingerprint((String) settings.get(VariableNames.FINGERPRINT_TYPE_NAME));
for (CombinedSingleCandidateMetFragProcess scmfp : processes) {
/*
* check whether the single run was successful
*/
if (scmfp.wasSuccessful()) {
ICandidate candidate = scmfp.getScoredPrecursorCandidates()[0];
java.util.ArrayList<MassFingerprintMatch> lossMatchlist = new java.util.ArrayList<MassFingerprintMatch>();
MatchList matchlist = candidate.getMatchList();
if (matchlist != null) {
candidate.initialisePrecursorCandidate();
for (int i = 0; i < matchlist.getNumberElements(); i++) {
IMatch matchI = matchlist.getElement(i);
IFragment fragmentI = matchI.getBestMatchedFragment();
double peakMassI = matchI.getMatchedPeak().getMass();
for (int j = i + 1; j < matchlist.getNumberElements(); j++) {
IMatch matchJ = matchlist.getElement(j);
double peakMassJ = matchJ.getMatchedPeak().getMass();
IFragment fragmentJ = matchJ.getBestMatchedFragment();
if (fragmentJ.isRealSubStructure(fragmentI)) {
double diff = MathTools.round(peakMassJ - peakMassI);
MassToFingerprintGroupList matchingLossToFingerprintGroupList = lossToFingerprintGroupListCollection.getElementByPeak(diff, mzppm, mzabs);
if (matchingLossToFingerprintGroupList == null)
continue;
IFragment diffFragment = fragmentJ.getDifferenceFragment(candidate.getPrecursorMolecule(), fragmentI);
if (diffFragment == null)
continue;
IAtomContainer con = fingerprint.getNormalizedAtomContainer(candidate.getPrecursorMolecule(), diffFragment);
lossMatchlist.add(new MassFingerprintMatch(diff, fingerprint.getNormalizedFastBitArrayFingerprint(con)));
}
}
// do the same for the precursor ion
double diff = MathTools.round(ionmass - peakMassI);
MassToFingerprintGroupList matchingLossToFingerprintGroupList = lossToFingerprintGroupListCollection.getElementByPeak(diff, mzppm, mzabs);
if (matchingLossToFingerprintGroupList == null)
continue;
IFragment diffFragment = fragmentI.getDifferenceFragment(candidate.getPrecursorMolecule());
if (diffFragment == null)
continue;
IAtomContainer con = fingerprint.getNormalizedAtomContainer(candidate.getPrecursorMolecule(), diffFragment);
lossMatchlist.add(new MassFingerprintMatch(diff, fingerprint.getNormalizedFastBitArrayFingerprint(con)));
}
}
// java.util.LinkedList<Double> nonExplainedLosses = this.getNonExplainedLoss(peakList, matchlist);
for (int j = 0; j < lossMatchlist.size(); j++) {
MassFingerprintMatch lossMatch = lossMatchlist.get(j);
MassToFingerprintGroupList lossToFingerprintGroupList = lossToFingerprintGroupListCollection.getElementByPeak(lossMatch.getMass(), mzppm, mzabs);
// if not loss not in our annotation list, there's no need to consider it
if (lossToFingerprintGroupList == null)
continue;
// lossMatch.setMass(lossToFingerprintGroupList.getPeakmz());
FastBitArray currentFingerprint = lossMatch.getFingerprint();
// check whether fingerprint was observed for current peak mass in the training data
if (!lossToFingerprintGroupList.containsFingerprint(currentFingerprint)) {
// if not add the fingerprint to background by addFingerprint function
// addFingerprint checks also whether fingerprint was already added
lossMassToFingerprints.addFingerprint(lossMatch.getMass(), currentFingerprint);
}
}
java.util.LinkedList<?> lossMassesFoundInPeakList = (java.util.LinkedList<?>) settings.get(VariableNames.LOSS_MASSES_FOUND_PEAKLIST_NAME);
// important! now add all losses not assigned by that candidates
// this is to equalize all loss match lists in length over all candidates
this.addNonExplainedLosses(lossMassesFoundInPeakList, lossMatchlist);
candidate.setProperty("LossMatchList", lossMatchlist);
}
}
// alpha
double alpha = (double) settings.get(VariableNames.LOSS_FINGERPRINT_ANNOTATION_ALPHA_VALUE_NAME);
// beta
double beta = (double) settings.get(VariableNames.LOSS_FINGERPRINT_ANNOTATION_BETA_VALUE_NAME);
// f_s
double f_seen_matched = (double) settings.get(VariableNames.LOSS_FINGERPRINT_MATCHED_TUPLE_COUNT_NAME);
// f_s
double f_seen_non_matched = (double) settings.get(VariableNames.LOSS_FINGERPRINT_NON_MATCHED_TUPLE_COUNT_NAME);
// f_u
double f_unseen_matched = lossMassToFingerprints.getOverallMatchedSize();
// f_u
double f_unseen_non_matched = lossMassToFingerprints.getOverallNonMatchedSize();
// \sum_N
double sumFingerprintFrequencies = (double) settings.get(VariableNames.LOSS_FINGERPRINT_DENOMINATOR_COUNT_NAME);
// set value for denominator of P(f,m)
double denominatorValue = sumFingerprintFrequencies + alpha * (f_seen_matched + f_unseen_matched) + beta * (f_seen_non_matched + f_unseen_non_matched);
settings.set(VariableNames.LOSS_FINGERPRINT_DENOMINATOR_VALUE_NAME, denominatorValue);
// P(f,m) F_u
double alphaProbability = alpha / denominatorValue;
// p(f,m) not annotated
double betaProbability = beta / denominatorValue;
for (int i = 0; i < lossToFingerprintGroupListCollection.getNumberElements(); i++) {
MassToFingerprintGroupList groupList = lossToFingerprintGroupListCollection.getElement(i);
// sum_f P(f,m)
// calculate sum of MF_s (including the alpha count) and the joint probabilities
// at this stage getProbability() returns the absolute counts from the annotation files
double sum_f = 0.0;
double sumFsProbabilities = 0.0;
for (int ii = 0; ii < groupList.getNumberElements(); ii++) {
// first calculate P(f,m)
if (groupList.getElement(ii).getFingerprint().getSize() != 1)
groupList.getElement(ii).setJointProbability((groupList.getElement(ii).getProbability() + alpha) / denominatorValue);
else
groupList.getElement(ii).setJointProbability((groupList.getElement(ii).getProbability() + beta) / denominatorValue);
// sum_f P(f,m) -> for F_s
sumFsProbabilities += groupList.getElement(ii).getJointProbability();
}
// calculate the sum of probabilities for un-observed fingerprints for the current mass
double sumFuProbabilities = alphaProbability * lossMassToFingerprints.getSizeMatched(groupList.getPeakmz());
// not needed as it's defined by fingerprint = "0"
// sumFuProbabilities += betaProbability * lossMassToFingerprints.getSizeNonMatched(groupList.getPeakmz());
sumFuProbabilities += betaProbability;
sum_f += sumFsProbabilities;
sum_f += sumFuProbabilities;
for (int ii = 0; ii < groupList.getNumberElements(); ii++) {
// second calculate P(f|m)
groupList.getElement(ii).setConditionalProbability_sp(groupList.getElement(ii).getJointProbability() / sum_f);
}
groupList.setAlphaProb(alphaProbability / sum_f);
groupList.setBetaProb(betaProbability / sum_f);
groupList.setProbabilityToConditionalProbability_sp();
groupList.calculateSumProbabilites();
}
return;
}
Aggregations