Examples with ModificationProvider com.compomics.util.experiment.biology.modifications.ModificationProvider used on opensource projects

Example 1 with ModificationProvider

use of com.compomics.util.experiment.biology.modifications.ModificationProvider in project peptide-shaker by compomics.

the class ModificationLocalizationScorer method scorePTMs.

/**
 * Scores PTMs in a protein match.
 *
 * @param identification The identification object containing the matches.
 * @param proteinMatch The protein match.
 * @param identificationParameters The identification parameters.
 * @param scorePeptides If true, peptides will be scored as well.
 * @param modificationProvider The modification provider to use.
 * @param waitingHandler The waiting handler to sue, ignored if null.
 */
public void scorePTMs(Identification identification, ProteinMatch proteinMatch, IdentificationParameters identificationParameters, boolean scorePeptides, ModificationProvider modificationProvider, WaitingHandler waitingHandler) {
    HashMap<Integer, ArrayList<String>> confidentSites = new HashMap<>();
    HashMap<Integer, HashMap<Integer, HashSet<String>>> ambiguousSites = new HashMap<>();
    for (long peptideKey : proteinMatch.getPeptideMatchesKeys()) {
        PeptideMatch peptideMatch = identification.getPeptideMatch(peptideKey);
        Peptide peptide = peptideMatch.getPeptide();
        PSParameter psParameter = (PSParameter) peptideMatch.getUrParam(PSParameter.dummy);
        if (psParameter.getMatchValidationLevel().isValidated() && peptide.getNVariableModifications() > 0) {
            PSModificationScores peptideScores = (PSModificationScores) peptideMatch.getUrParam(PSModificationScores.dummy);
            if (peptideScores == null || scorePeptides) {
                scorePTMs(identification, peptideMatch, identificationParameters, modificationProvider, waitingHandler);
                peptideScores = (PSModificationScores) peptideMatch.getUrParam(PSModificationScores.dummy);
            }
            if (peptideScores != null) {
                int[] peptideStart = peptide.getProteinMapping().get(proteinMatch.getLeadingAccession());
                for (int confidentSite : peptideScores.getConfidentSites()) {
                    for (int peptideTempStart : peptideStart) {
                        int siteOnProtein = peptideTempStart + confidentSite - 1;
                        ArrayList<String> modificationsAtSite = confidentSites.get(siteOnProtein);
                        if (modificationsAtSite == null) {
                            modificationsAtSite = new ArrayList<>();
                            confidentSites.put(siteOnProtein, modificationsAtSite);
                        }
                        for (String modName : peptideScores.getConfidentModificationsAt(confidentSite)) {
                            if (!modificationsAtSite.contains(modName)) {
                                modificationsAtSite.add(modName);
                            }
                        }
                    }
                }
                for (int representativeSite : peptideScores.getRepresentativeSites()) {
                    HashMap<Integer, HashSet<String>> peptideAmbiguousSites = peptideScores.getAmbiguousModificationsAtRepresentativeSite(representativeSite);
                    for (int peptideTempStart : peptideStart) {
                        int proteinRepresentativeSite = peptideTempStart + representativeSite - 1;
                        HashMap<Integer, HashSet<String>> proteinAmbiguousSites = ambiguousSites.get(proteinRepresentativeSite);
                        if (proteinAmbiguousSites == null) {
                            proteinAmbiguousSites = new HashMap<>(peptideAmbiguousSites.size());
                            ambiguousSites.put(proteinRepresentativeSite, proteinAmbiguousSites);
                        }
                        for (int peptideSite : peptideAmbiguousSites.keySet()) {
                            int siteOnProtein = peptideTempStart + peptideSite - 1;
                            proteinAmbiguousSites.put(siteOnProtein, peptideAmbiguousSites.get(peptideSite));
                        }
                    }
                }
            }
        }
    }
    // remove ambiguous sites where a confident was found and merge overlapping groups
    PSModificationScores proteinScores = new PSModificationScores();
    ArrayList<Integer> representativeSites = new ArrayList<>(ambiguousSites.keySet());
    Collections.sort(representativeSites);
    for (Integer representativeSite : representativeSites) {
        HashMap<Integer, HashSet<String>> secondarySitesMap = ambiguousSites.get(representativeSite);
        ArrayList<Integer> secondarySites = new ArrayList<>(secondarySitesMap.keySet());
        for (int secondarySite : secondarySites) {
            ArrayList<String> confidentModifications = confidentSites.get(secondarySite);
            if (confidentModifications != null) {
                boolean sameModification = confidentModifications.stream().map(modName -> modificationProvider.getModification(modName)).anyMatch(confidentModification -> secondarySitesMap.get(secondarySite).stream().map(modName -> modificationProvider.getModification(modName)).anyMatch(secondaryModification -> secondaryModification.getMass() == confidentModification.getMass()));
                if (sameModification) {
                    ambiguousSites.remove(representativeSite);
                    break;
                }
            }
            if (secondarySite != representativeSite) {
                ArrayList<Integer> tempRepresentativeSites = new ArrayList<>(ambiguousSites.keySet());
                Collections.sort(tempRepresentativeSites);
                for (Integer previousSite : tempRepresentativeSites) {
                    if (previousSite >= representativeSite) {
                        break;
                    }
                    if (previousSite == secondarySite) {
                        HashMap<Integer, HashSet<String>> previousSites = ambiguousSites.get(previousSite);
                        HashSet<String> previousModifications = previousSites.get(previousSite);
                        boolean sameModification = previousModifications.stream().map(modName -> modificationProvider.getModification(modName)).anyMatch(previousModification -> secondarySitesMap.get(secondarySite).stream().map(modName -> modificationProvider.getModification(modName)).anyMatch(secondaryModification -> secondaryModification.getMass() == previousModification.getMass()));
                        if (sameModification) {
                            for (int tempSecondarySite : secondarySitesMap.keySet()) {
                                if (!previousSites.containsKey(secondarySite)) {
                                    previousSites.put(tempSecondarySite, secondarySitesMap.get(tempSecondarySite));
                                }
                            }
                            ambiguousSites.remove(representativeSite);
                        }
                    }
                }
            }
        }
    }
    for (int confidentSite : confidentSites.keySet()) {
        for (String modName : confidentSites.get(confidentSite)) {
            proteinScores.addConfidentModificationSite(modName, confidentSite);
        }
    }
    for (int representativeSite : ambiguousSites.keySet()) {
        proteinScores.addAmbiguousModificationSites(representativeSite, ambiguousSites.get(representativeSite));
    }
    proteinMatch.addUrParam(proteinScores);
}

Example 2 with ModificationProvider

use of com.compomics.util.experiment.biology.modifications.ModificationProvider in project peptide-shaker by compomics.

the class ModificationLocalizationScorer method attachProbabilisticScore.

/**
 * Attaches the selected probabilistic modification score.
 *
 * @param spectrumMatch The spectrum match studied, the score will be
 * calculated for the best assumption only.
 * @param sequenceProvider The protein sequence provider to use.
 * @param spectrumProvider The spectrum provider to use.
 * @param modificationProvider The modification provider to use.
 * @param identificationParameters The identification parameters.
 * @param peptideSpectrumAnnotator The peptide spectrum annotator to use.
 * @param identification The identification object containing the matches.
 */
private void attachProbabilisticScore(SpectrumMatch spectrumMatch, SequenceProvider sequenceProvider, SpectrumProvider spectrumProvider, ModificationProvider modificationProvider, IdentificationParameters identificationParameters, PeptideSpectrumAnnotator peptideSpectrumAnnotator, Identification identification) {
    SearchParameters searchParameters = identificationParameters.getSearchParameters();
    AnnotationParameters annotationParameters = identificationParameters.getAnnotationParameters();
    ModificationLocalizationParameters scoringParameters = identificationParameters.getModificationLocalizationParameters();
    SequenceMatchingParameters sequenceMatchingParameters = identificationParameters.getSequenceMatchingParameters();
    SequenceMatchingParameters modificationSequenceMatchingParameters = scoringParameters.getSequenceMatchingParameters();
    ModificationParameters modificationParameters = searchParameters.getModificationParameters();
    PSModificationScores modificationScores = (PSModificationScores) spectrumMatch.getUrParam(PSModificationScores.dummy);
    if (modificationScores != null) {
        modificationScores = new PSModificationScores();
        spectrumMatch.addUrParam(modificationScores);
    }
    HashMap<Double, ArrayList<Modification>> modificationsMap = new HashMap<>(1);
    HashMap<Double, Integer> nMod = new HashMap<>(1);
    PeptideAssumption bestPeptideAssumption = spectrumMatch.getBestPeptideAssumption();
    Peptide peptide = bestPeptideAssumption.getPeptide();
    for (ModificationMatch modificationMatch : peptide.getVariableModifications()) {
        Modification refMod = modificationProvider.getModification(modificationMatch.getModification());
        double modMass = refMod.getMass();
        if (!modificationsMap.containsKey(modMass)) {
            ArrayList<Modification> modifications = modificationFactory.getSameMassNotFixedModifications(modMass, searchParameters).stream().map(modification -> modificationProvider.getModification(modification)).collect(Collectors.toCollection(ArrayList::new));
            modificationsMap.put(modMass, modifications);
            nMod.put(modMass, 1);
        } else {
            nMod.put(modMass, nMod.get(modMass) + 1);
        }
    }
    if (!modificationsMap.isEmpty()) {
        String spectrumFile = spectrumMatch.getSpectrumFile();
        String spectrumTitle = spectrumMatch.getSpectrumTitle();
        Spectrum spectrum = spectrumProvider.getSpectrum(spectrumFile, spectrumTitle);
        SpecificAnnotationParameters specificAnnotationParameters = annotationParameters.getSpecificAnnotationParameters(spectrumFile, spectrumTitle, bestPeptideAssumption, modificationParameters, sequenceProvider, modificationSequenceMatchingParameters, peptideSpectrumAnnotator);
        for (double modMass : modificationsMap.keySet()) {
            HashMap<Integer, Double> scores = null;
            if (scoringParameters.getSelectedProbabilisticScore() == ModificationLocalizationScore.PhosphoRS) {
                scores = PhosphoRS.getSequenceProbabilities(peptide, modificationsMap.get(modMass), modificationParameters, spectrum, sequenceProvider, annotationParameters, specificAnnotationParameters, scoringParameters.isProbabilisticScoreNeutralLosses(), sequenceMatchingParameters, modificationSequenceMatchingParameters, peptideSpectrumAnnotator);
                if (scores == null) {
                    throw new IllegalArgumentException("An error occurred while scoring spectrum " + spectrumTitle + " of file " + spectrumFile + " with PhosphoRS.");
                // Most likely a compatibility issue with utilities
                }
            }
            if (scores != null) {
                // remap to searched modifications
                Modification mappedModification = null;
                String peptideSequence = peptide.getSequence();
                for (int site : scores.keySet()) {
                    if (site == 0) {
                        // N-term mod
                        for (Modification modification : modificationsMap.get(modMass)) {
                            if (modification.getModificationType().isNTerm()) {
                                mappedModification = modification;
                                break;
                            }
                        }
                        if (mappedModification == null) {
                            throw new IllegalArgumentException("Could not map the PTM of mass " + modMass + " on the N-terminus of the peptide " + peptideSequence + ".");
                        }
                    } else if (site == peptideSequence.length() + 1) {
                        // C-term mod
                        for (Modification modification : modificationsMap.get(modMass)) {
                            if (modification.getModificationType().isCTerm()) {
                                mappedModification = modification;
                                break;
                            }
                        }
                        if (mappedModification == null) {
                            throw new IllegalArgumentException("Could not map the PTM of mass " + modMass + " on the C-terminus of the peptide " + peptideSequence + ".");
                        }
                    } else {
                        for (Modification modification : modificationsMap.get(modMass)) {
                            mappedModification = modification;
                            break;
                        }
                        if (mappedModification == null) {
                            throw new IllegalArgumentException("Could not map the PTM of mass " + modMass + " at site " + site + " in peptide " + peptide.getSequence() + ".");
                        }
                    }
                    String modName = mappedModification.getName();
                    ModificationScoring modificationScoring = modificationScores.getModificationScoring(modName);
                    if (modificationScoring == null) {
                        modificationScoring = new ModificationScoring(modName);
                        modificationScores.addModificationScoring(modName, modificationScoring);
                    }
                    modificationScoring.setProbabilisticScore(site, scores.get(site));
                }
            }
        }
        identification.updateObject(spectrumMatch.getKey(), spectrumMatch);
    }
}

Example 3 with ModificationProvider

use of com.compomics.util.experiment.biology.modifications.ModificationProvider in project peptide-shaker by compomics.

the class ProteinProcessor method processProteins.

/**
 * Scores the PTMs of all protein matches contained in an identification
 * object, estimates spectrum counting and summary statistics.
 *
 * @param modificationLocalizationScorer The modification localization
 * scorer to use.
 * @param metrics If provided, metrics on proteins will be saved while
 * iterating the matches.
 * @param modificationProvider The modification provider to use.
 * @param waitingHandler The handler displaying feedback to the user.
 * @param exceptionHandler The exception handler to use.
 * @param processingParameters The processing parameters.
 *
 * @throws java.lang.InterruptedException exception thrown if a thread gets
 * interrupted
 * @throws java.util.concurrent.TimeoutException exception thrown if the
 * operation times out
 */
public void processProteins(ModificationLocalizationScorer modificationLocalizationScorer, Metrics metrics, ModificationProvider modificationProvider, WaitingHandler waitingHandler, ExceptionHandler exceptionHandler, ProcessingParameters processingParameters) throws InterruptedException, TimeoutException {
    waitingHandler.setWaitingText("Scoring Protein Modification Localization. Please Wait...");
    int max = identification.getProteinIdentification().size();
    waitingHandler.setSecondaryProgressCounterIndeterminate(false);
    waitingHandler.setMaxSecondaryProgressCounter(max);
    // validate the proteins
    ExecutorService pool = Executors.newFixedThreadPool(processingParameters.getnThreads());
    ProteinMatchesIterator proteinMatchesIterator = identification.getProteinMatchesIterator(waitingHandler);
    ArrayList<ProteinRunnable> runnables = new ArrayList<>(processingParameters.getnThreads());
    for (int i = 1; i <= processingParameters.getnThreads(); i++) {
        ProteinRunnable runnable = new ProteinRunnable(proteinMatchesIterator, modificationLocalizationScorer, modificationProvider, waitingHandler, exceptionHandler);
        pool.submit(runnable);
        runnables.add(runnable);
    }
    if (waitingHandler.isRunCanceled()) {
        pool.shutdownNow();
    }
    pool.shutdown();
    if (!pool.awaitTermination(identification.getProteinIdentification().size(), TimeUnit.MINUTES)) {
        throw new InterruptedException("Protein matches validation timed out. Please contact the developers.");
    }
    waitingHandler.setSecondaryProgressCounterIndeterminate(true);
    if (metrics != null) {
        metrics.setMaxSpectrumCounting(runnables.stream().mapToDouble(ProteinRunnable::getMaxSpectrumCounting).sum());
        metrics.setnValidatedProteins(runnables.stream().mapToInt(ProteinRunnable::getnValidatedProteins).sum());
        metrics.setnConfidentProteins(runnables.stream().mapToInt(ProteinRunnable::getnConfidentProteins).sum());
        metrics.setMaxNPeptides(runnables.stream().mapToInt(ProteinRunnable::getMaxPeptides).max().orElse(0));
        metrics.setMaxNPsms(runnables.stream().mapToInt(ProteinRunnable::getMaxPsms).max().orElse(0));
        metrics.setMaxMW(runnables.stream().mapToDouble(ProteinRunnable::getMaxMW).max().orElse(0.0));
        metrics.setMaxProteinAccessionLength(runnables.stream().mapToInt(ProteinRunnable::getMaxProteinAccessionLength).max().orElse(0));
        TreeMap<Double, TreeMap<Integer, TreeMap<Integer, TreeSet<Long>>>> orderMap1 = new TreeMap<>();
        for (int i = 0; i < runnables.size(); i++) {
            HashMap<Double, HashMap<Integer, HashMap<Integer, HashSet<Long>>>> threadMap1 = runnables.get(i).getOrderMap();
            for (Entry<Double, HashMap<Integer, HashMap<Integer, HashSet<Long>>>> entry1 : threadMap1.entrySet()) {
                double key1 = entry1.getKey();
                HashMap<Integer, HashMap<Integer, HashSet<Long>>> threadMap2 = entry1.getValue();
                TreeMap<Integer, TreeMap<Integer, TreeSet<Long>>> orderMap2 = orderMap1.get(key1);
                if (orderMap2 == null) {
                    orderMap2 = new TreeMap<>();
                    orderMap1.put(key1, orderMap2);
                }
                for (Entry<Integer, HashMap<Integer, HashSet<Long>>> entry2 : threadMap2.entrySet()) {
                    int key2 = entry2.getKey();
                    HashMap<Integer, HashSet<Long>> threadMap3 = entry2.getValue();
                    TreeMap<Integer, TreeSet<Long>> orderMap3 = orderMap2.get(key2);
                    if (orderMap3 == null) {
                        orderMap3 = new TreeMap<>();
                        orderMap2.put(key2, orderMap3);
                    }
                    for (Entry<Integer, HashSet<Long>> entry3 : threadMap3.entrySet()) {
                        int key3 = entry3.getKey();
                        HashSet<Long> threadSet = entry3.getValue();
                        TreeSet<Long> orderedSet = orderMap3.get(key3);
                        if (orderedSet == null) {
                            orderedSet = new TreeSet<>();
                            orderMap3.put(key3, orderedSet);
                        }
                        orderedSet.addAll(threadSet);
                    }
                }
            }
        }
        long[] proteinKeys = orderMap1.values().stream().flatMap(map -> map.values().stream()).flatMap(map -> map.values().stream()).flatMap(set -> set.stream()).mapToLong(a -> a).toArray();
        metrics.setProteinKeys(proteinKeys);
    }
}