Examples with FastaSummary com.compomics.util.experiment.io.biology.protein.FastaSummary used on opensource projects

Example 1 with FastaSummary

use of com.compomics.util.experiment.io.biology.protein.FastaSummary in project peptide-shaker by compomics.

the class PeptideShakerGUI method importPeptideShakerFile.

/**
 * Imports information from a PeptideShaker file.
 *
 * @param psFile The PeptideShaker file to import.
 * @param importFromZip flag that determines if psdb was imported from a zip
 * file
 */
public void importPeptideShakerFile(File psFile, boolean importFromZip) {
    psdbParent.setPsdbFile(psFile);
    psdbParent.setPsdbImportFromZip(importFromZip);
    // needed due to threading issues
    final PeptideShakerGUI peptideShakerGUI = this;
    progressDialog = new ProgressDialogX(this, Toolkit.getDefaultToolkit().getImage(getClass().getResource("/icons/peptide-shaker.gif")), Toolkit.getDefaultToolkit().getImage(getClass().getResource("/icons/peptide-shaker-orange.gif")), true);
    progressDialog.setPrimaryProgressCounterIndeterminate(true);
    progressDialog.setTitle("Importing Project. Please Wait...");
    // reset the title
    resetFrameTitle();
    new Thread(new Runnable() {

        public void run() {
            try {
                progressDialog.setVisible(true);
            } catch (IndexOutOfBoundsException e) {
            // ignore
            }
        }
    }, "ProgressDialog").start();
    new Thread("ImportThread") {

        @Override
        public void run() {
            try {
                // reset enzymes, modifications and parameters
                resetModificationFactory();
                setDefaultParameters();
                setCurentNotes(new ArrayList<String>(0));
                updateNotesNotificationCounter();
                openingExistingProject = true;
                psdbParent.loadPsdbFile(PeptideShaker.getMatchesFolder(), progressDialog, psdbParent.getPsdbImportFromZip());
                // load project specific PTMs
                String error = PeptideShaker.loadModifications(getIdentificationParameters().getSearchParameters());
                if (error != null) {
                    JOptionPane.showMessageDialog(peptideShakerGUI, error, "Modification Definition Changed", JOptionPane.WARNING_MESSAGE);
                }
                // resets the display features generator according to the new project
                resetDisplayFeaturesGenerator();
                if (progressDialog.isRunCanceled()) {
                    clearData(true, true);
                    clearParameters();
                    progressDialog.setRunFinished();
                    openingExistingProject = false;
                    return;
                }
                progressDialog.setTitle("Loading FASTA File. Please Wait...");
                boolean fileFound = true;
                try {
                    FastaSummary fastaSummary = psdbParent.loadFastaFile(progressDialog);
                    if (fastaSummary == null) {
                        fileFound = false;
                    }
                } catch (IOException e) {
                    fileFound = false;
                }
                if (!fileFound && !locateFastaFileManually()) {
                    String fastaFilePath = getProjectDetails().getFastaFile();
                    JOptionPane.showMessageDialog(peptideShakerGUI, "An error occurred while reading:\n" + fastaFilePath + "." + "\n\nFile not found.", "File Input Error", JOptionPane.ERROR_MESSAGE);
                    clearData(true, true);
                    clearParameters();
                    progressDialog.setRunFinished();
                    openingExistingProject = false;
                    return;
                }
                if (progressDialog.isRunCanceled()) {
                    clearData(true, true);
                    clearParameters();
                    progressDialog.setRunFinished();
                    openingExistingProject = false;
                    return;
                }
                progressDialog.setTitle("Loading Spectrum Files. Please Wait...");
                progressDialog.resetPrimaryProgressCounter();
                progressDialog.setMaxPrimaryProgressCounter(getIdentification().getSpectrumIdentification().size() + 1);
                progressDialog.increasePrimaryProgressCounter();
                ProjectDetails projectParameters = psdbParent.getProjectDetails();
                Set<String> fileNames = getProjectDetails().getSpectrumFileNames();
                ArrayList<File> spectrumFiles = fileNames.stream().map(projectParameters::getSpectrumFilePath).map(path -> new File(path)).collect(Collectors.toCollection(ArrayList::new));
                int cpt = 0, total = fileNames.size();
                for (String spectrumFileName : getIdentification().getFractions()) {
                    progressDialog.setTitle("Loading Spectrum Files (" + ++cpt + " of " + total + "). Please Wait...");
                    progressDialog.increasePrimaryProgressCounter();
                    boolean found;
                    try {
                        found = psdbParent.loadSpectrumFile(spectrumFileName, spectrumFiles, progressDialog);
                    } catch (Exception e) {
                        found = false;
                    }
                    if (!found) {
                        JOptionPane.showMessageDialog(peptideShakerGUI, "Spectrum file not found: \'" + spectrumFileName + "\'." + "\nPlease select the spectrum file or the folder containing it manually.", "File Not Found", JOptionPane.WARNING_MESSAGE);
                        JFileChooser fileChooser = new JFileChooser(getLastSelectedFolder().getLastSelectedFolder());
                        fileChooser.setDialogTitle("Open Spectrum File");
                        FileFilter filter = new FileFilter() {

                            @Override
                            public boolean accept(File myFile) {
                                return myFile.getName().toLowerCase().endsWith(".mgf") || myFile.getName().toLowerCase().endsWith(".mgf.gz") || myFile.getName().toLowerCase().endsWith(".mzml") || myFile.getName().toLowerCase().endsWith(".mzml.gz") || myFile.isDirectory();
                            }

                            @Override
                            public String getDescription() {
                                return "Supported formats: mgf or mzML (.mgf, .mg.gz, .mzml, .mzml.gz)";
                            }
                        };
                        fileChooser.setFileFilter(filter);
                        int returnVal = fileChooser.showDialog(peptideShakerGUI, "Open");
                        if (returnVal == JFileChooser.APPROVE_OPTION) {
                            File spectrumFolder = fileChooser.getSelectedFile();
                            if (!spectrumFolder.isDirectory()) {
                                spectrumFolder = spectrumFolder.getParentFile();
                            }
                            lastSelectedFolder.setLastSelectedFolder(spectrumFolder.getAbsolutePath());
                            found = false;
                            for (File file : spectrumFolder.listFiles()) {
                                for (String spectrumFileName2 : getIdentification().getFractions()) {
                                    try {
                                        String fileName = IoUtil.removeExtension(file.getName());
                                        if (spectrumFileName2.equals(fileName)) {
                                            psdbParent.loadSpectrumFile(file, progressDialog);
                                            spectrumFiles.add(file);
                                        }
                                        if (fileName.equals(spectrumFileName2)) {
                                            found = true;
                                        }
                                    } catch (Exception e) {
                                    // ignore
                                    }
                                }
                            }
                            if (found) {
                                try {
                                    found = psdbParent.loadSpectrumFile(spectrumFileName, spectrumFiles, progressDialog);
                                } catch (Exception e) {
                                    found = false;
                                }
                            }
                            if (!found) {
                                JOptionPane.showMessageDialog(peptideShakerGUI, spectrumFileName + " was not found in the given folder.", "File Input Error", JOptionPane.ERROR_MESSAGE);
                                clearData(true, true);
                                clearParameters();
                                progressDialog.setRunFinished();
                                openingExistingProject = false;
                                return;
                            }
                        }
                    }
                    if (progressDialog.isRunCanceled()) {
                        clearData(true, true);
                        clearParameters();
                        progressDialog.setRunFinished();
                        openingExistingProject = false;
                        return;
                    }
                }
                progressDialog.setPrimaryProgressCounterIndeterminate(true);
                progressDialog.setRunFinished();
                peptideShakerGUI.displayResults();
                // display the overview tab data
                allTabsJTabbedPaneStateChanged(null);
                peptideShakerGUI.updateFrameTitle();
                dataSaved = true;
                SwingUtilities.invokeLater(new Runnable() {

                    public void run() {
                        openingExistingProject = false;
                    }
                });
            } catch (OutOfMemoryError error) {
                System.err.println("Ran out of memory!");
                System.err.println("Memory given to the Java virtual machine: " + Runtime.getRuntime().maxMemory() + ".");
                System.err.println("Memory used by the Java virtual machine: " + Runtime.getRuntime().totalMemory() + ".");
                System.err.println("Free memory in the Java virtual machine: " + Runtime.getRuntime().freeMemory() + ".");
                Runtime.getRuntime().gc();
                String message = "PeptideShaker used up all the available memory and had to be stopped.<br>" + "Memory boundaries are changed in the the Welcome Dialog (Settings<br>" + "& Help > Settings > Java Memory Settings) or in the Edit menu (Edit<br>" + "Java Options). See also <a href=\"https://compomics.github.io/projects/compomics-utilities/wiki/JavaTroubleShooting.html\">JavaTroubleShooting</a>.";
                JOptionPane.showMessageDialog(PeptideShakerGUI.this, JOptionEditorPane.getJOptionEditorPane(message), "Out Of Memory", JOptionPane.ERROR_MESSAGE);
                progressDialog.setRunFinished();
                error.printStackTrace();
            } catch (OptionalDataException e) {
                progressDialog.setRunFinished();
                if (e.eof) {
                    JOptionPane.showMessageDialog(peptideShakerGUI, "An error occurred while reading:\n" + psdbParent.getPsdbFile() + ".\n\n" + "The end of the file was reached unexpectedly. The file seems to be corrupt and cannot\n" + "be opened. If the file is a copy, make sure that it is identical to the original file.", "File Input Error", JOptionPane.ERROR_MESSAGE);
                } else {
                    JOptionPane.showMessageDialog(peptideShakerGUI, "An error occurred while reading:\n" + psdbParent.getPsdbFile() + ".\n\n" + "Please verify that the version used to create the file\n" + "is compatible with your version of PeptideShaker.", "File Input Error", JOptionPane.ERROR_MESSAGE);
                }
                e.printStackTrace();
            } catch (EOFException e) {
                progressDialog.setRunFinished();
                JOptionPane.showMessageDialog(peptideShakerGUI, "An error occurred while reading:\n" + psdbParent.getPsdbFile() + ".\n\n" + "The end of the file was reached unexpectedly. The file seems to be corrupt and cannot\n" + "be opened. If the file is a copy, make sure that it is identical to the original file.", "File Input Error", JOptionPane.ERROR_MESSAGE);
                e.printStackTrace();
            } catch (Exception e) {
                progressDialog.setRunFinished();
                JOptionPane.showMessageDialog(peptideShakerGUI, "An error occurred while reading:\n" + psdbParent.getPsdbFile() + ".\n\n" + "Please verify that the version used to create the file\n" + "is compatible with your version of PeptideShaker.", "File Input Error", JOptionPane.ERROR_MESSAGE);
                e.printStackTrace();
            }
        }
    }.start();
}

Example 2 with FastaSummary

use of com.compomics.util.experiment.io.biology.protein.FastaSummary in project peptide-shaker by compomics.

the class MzIdentMLExportDialog method convertJButtonActionPerformed.

// GEN-LAST:event_contactEmailJTextFieldKeyReleased
/**
 * Convert the project to an mzIdentML file.
 *
 * @param evt
 */
private void convertJButtonActionPerformed(java.awt.event.ActionEvent evt) {
    // GEN-FIRST:event_convertJButtonActionPerformed
    final File finalOutputFile = new File(outputFolderJTextField.getText());
    progressDialog = new ProgressDialogX(peptideShakerGUI, Toolkit.getDefaultToolkit().getImage(getClass().getResource("/icons/peptide-shaker.gif")), Toolkit.getDefaultToolkit().getImage(getClass().getResource("/icons/peptide-shaker-orange.gif")), true);
    progressDialog.setPrimaryProgressCounterIndeterminate(true);
    progressDialog.setTitle("Exporting mzIdentML. Please Wait...");
    new Thread(new Runnable() {

        public void run() {
            try {
                progressDialog.setVisible(true);
            } catch (IndexOutOfBoundsException e) {
            // ignore
            }
        }
    }, "ProgressDialog").start();
    new Thread("ConvertThread") {

        @Override
        public void run() {
            IdentificationParameters identificationParameters = peptideShakerGUI.getIdentificationParameters();
            AnnotationParameters annotationParameters = identificationParameters.getAnnotationParameters();
            ProjectDetails projectDetails = peptideShakerGUI.getProjectDetails();
            // save the inserted mzid details with the project
            projectDetails.setContactFirstName(contactFirstNameJTextField.getText().trim());
            projectDetails.setContactLastName(contactLastNameJTextField.getText().trim());
            projectDetails.setContactEmail(contactEmailJTextField.getText().trim());
            projectDetails.setContactAddress(contactAddressJTextField.getText().trim());
            if (!contactUrlJTextField.getText().trim().isEmpty()) {
                projectDetails.setContactUrl(contactUrlJTextField.getText().trim());
            } else {
                projectDetails.setContactUrl(null);
            }
            projectDetails.setOrganizationName(organizationNameJTextField.getText().trim());
            projectDetails.setOrganizationEmail(organizationEmailJTextField.getText().trim());
            projectDetails.setOrganizationAddress(organizationAddressJTextField.getText().trim());
            if (!organizationUrlJTextField.getText().trim().isEmpty()) {
                projectDetails.setOrganizationUrl(organizationUrlJTextField.getText().trim());
            } else {
                projectDetails.setOrganizationUrl(null);
            }
            projectDetails.setIncludeProteinSequences(includeSequencesCheckBox.isSelected());
            projectDetails.setMzIdentOutputFile(outputFolderJTextField.getText());
            peptideShakerGUI.setDataSaved(false);
            boolean conversionCompleted = false;
            // make sure that all annotations are included
            double currentIntensityLimit = annotationParameters.getAnnotationIntensityLimit();
            annotationParameters.setIntensityLimit(0.0);
            try {
                FastaSummary fastaSummary = FastaSummary.getSummary(projectDetails.getFastaFile(), identificationParameters.getFastaParameters(), progressDialog);
                MzIdentMLExport mzIdentMLExport = new MzIdentMLExport(PeptideShaker.getVersion(), peptideShakerGUI.getIdentification(), projectDetails, identificationParameters, peptideShakerGUI.getSequenceProvider(), peptideShakerGUI.getProteinDetailsProvider(), peptideShakerGUI.getSpectrumProvider(), ModificationFactory.getInstance(), fastaSummary, peptideShakerGUI.getIdentificationFeaturesGenerator(), finalOutputFile, includeSequencesCheckBox.isSelected(), progressDialog, true);
                mzIdentMLExport.createMzIdentMLFile(mzIdentMLVersion);
                // validate the mzidentml file
                if (validateMzIdentML && !progressDialog.isRunCanceled()) {
                    progressDialog.setPrimaryProgressCounterIndeterminate(true);
                    progressDialog.setTitle("Validating mzIdentML. Please Wait...");
                    String errors = validateMzIdentML(finalOutputFile);
                    // see if any errors were found, and display them to the user
                    if (!errors.isEmpty()) {
                        JOptionPane.showMessageDialog(null, errors, "mzIdentML Errors", JOptionPane.ERROR_MESSAGE);
                    } else {
                        conversionCompleted = true;
                    }
                } else {
                    conversionCompleted = true;
                }
            } catch (Exception e) {
                peptideShakerGUI.catchException(e);
                progressDialog.setRunCanceled();
                progressDialog.dispose();
                return;
            } finally {
                // reset the annotation level
                annotationParameters.setIntensityLimit(currentIntensityLimit);
            }
            // close the progress dialog
            boolean processCancelled = progressDialog.isRunCanceled();
            progressDialog.setRunFinished();
            // display a conversion complete message to the user
            if (conversionCompleted && !processCancelled) {
                JOptionPane.showMessageDialog(MzIdentMLExportDialog.this, JOptionEditorPane.getJOptionEditorPane("mzIdentML file \'" + new File(outputFolderJTextField.getText()).getAbsolutePath() + "\' created." + "<br><br>" + "Review your mzIdentML files with <a href=\"https://github.com/PRIDE-Toolsuite/pride-inspector\">PRIDE Inspector</a>.<br>" + "Publish your mzIdentML files via <a href=\"http://www.proteomexchange.org/submission\">ProteomeXchange</a>."), "File Created", JOptionPane.INFORMATION_MESSAGE);
                dispose();
            }
            if (processCancelled) {
                JOptionPane.showMessageDialog(peptideShakerGUI, "mzIdentML conversion cancelled by the user.", "mzIdentML Conversion Cancelled", JOptionPane.WARNING_MESSAGE);
            }
        }
    }.start();
}

Example 3 with FastaSummary

use of com.compomics.util.experiment.io.biology.protein.FastaSummary in project peptide-shaker by compomics.

the class PeptideShaker method createProject.

/**
 * Creates a PeptideShaker project.
 *
 * @param waitingHandler the handler displaying feedback to the user
 * @param exceptionHandler handler for exceptions
 * @param identificationParameters the identification parameters
 * @param processingParameters the processing parameters
 * @param projectType the project type
 * @param spectrumCountingParameters the spectrum counting parameters
 * @param spectrumProvider the spectrum provider
 * @param setWaitingHandlerFinshedWhenDone if true, the waiting handler is
 * set to finished when the project is created
 * @param projectDetails the project details
 *
 * @throws java.lang.InterruptedException exception thrown if a thread gets
 * interrupted
 * @throws java.util.concurrent.TimeoutException exception thrown if a
 * process times out
 * @throws java.io.IOException if an exception occurs when parsing files
 */
public void createProject(IdentificationParameters identificationParameters, ProcessingParameters processingParameters, SpectrumCountingParameters spectrumCountingParameters, SpectrumProvider spectrumProvider, ProjectDetails projectDetails, ProjectType projectType, WaitingHandler waitingHandler, boolean setWaitingHandlerFinshedWhenDone, ExceptionHandler exceptionHandler) throws InterruptedException, TimeoutException, IOException {
    identification.getObjectsDB().commit();
    identificationFeaturesGenerator = new IdentificationFeaturesGenerator(identification, identificationParameters, sequenceProvider, spectrumProvider, metrics, spectrumCountingParameters);
    matchesValidator = new MatchesValidator(new TargetDecoyMap(), new TargetDecoyMap(), new TargetDecoyMap());
    if (waitingHandler.isRunCanceled()) {
        return;
    }
    PsmScoringParameters psmScoringPreferences = identificationParameters.getPsmScoringParameters();
    FastaParameters fastaParameters = identificationParameters.getFastaParameters();
    FastaSummary fastaSummary = FastaSummary.getSummary(projectDetails.getFastaFile(), fastaParameters, waitingHandler);
    // set the background species
    identificationParameters.getGeneParameters().setBackgroundSpeciesFromFastaSummary(fastaSummary);
    ArrayList<Integer> usedAlgorithms = projectDetails.getIdentificationAlgorithms();
    if (psmScoringPreferences.isScoringNeeded(usedAlgorithms)) {
        waitingHandler.appendReport("Estimating PSM scores.", true, true);
        PsmScorer psmScorer = new PsmScorer(fastaParameters, sequenceProvider, spectrumProvider);
        psmScorer.estimateIntermediateScores(identification, inputMap, processingParameters, identificationParameters, waitingHandler, exceptionHandler);
        if (psmScoringPreferences.isTargetDecoyNeededForPsmScoring(usedAlgorithms)) {
            if (fastaParameters.isTargetDecoy()) {
                waitingHandler.appendReport("Estimating intermediate scores probabilities.", true, true);
                psmScorer.estimateIntermediateScoreProbabilities(identification, inputMap, processingParameters, waitingHandler);
            } else {
                waitingHandler.appendReport("No decoy sequences found. Impossible to " + "estimate intermediate scores probabilities.", true, true);
            }
        }
        waitingHandler.appendReport("Scoring PSMs.", true, true);
        psmScorer.scorePsms(identification, inputMap, processingParameters, identificationParameters, waitingHandler);
    }
    identification.getObjectsDB().commit();
    System.gc();
    if (fastaParameters.isTargetDecoy()) {
        waitingHandler.appendReport("Computing assumptions probabilities.", true, true);
    } else {
        waitingHandler.appendReport("Importing assumptions scores.", true, true);
    }
    inputMap.estimateProbabilities(waitingHandler);
    waitingHandler.increasePrimaryProgressCounter();
    if (waitingHandler.isRunCanceled()) {
        return;
    }
    identification.getObjectsDB().commit();
    System.gc();
    waitingHandler.appendReport("Saving assumptions probabilities, selecting best match, scoring modification localization.", true, true);
    PsmProcessor psmProcessor = new PsmProcessor(identification);
    psmProcessor.processPsms(inputMap, identificationParameters, matchesValidator, modificationLocalizationScorer, sequenceProvider, spectrumProvider, modificationFactory, proteinCount, processingParameters.getnThreads(), waitingHandler, exceptionHandler);
    waitingHandler.increasePrimaryProgressCounter();
    if (waitingHandler.isRunCanceled()) {
        return;
    }
    identification.getObjectsDB().commit();
    System.gc();
    waitingHandler.appendReport("Computing PSM probabilities.", true, true);
    matchesValidator.getPsmMap().estimateProbabilities(waitingHandler);
    if (waitingHandler.isRunCanceled()) {
        return;
    }
    identification.getObjectsDB().commit();
    System.gc();
    if (projectType == ProjectType.peptide || projectType == ProjectType.protein) {
        PeptideInference peptideInference = new PeptideInference();
        ModificationLocalizationParameters modificationScoringPreferences = identificationParameters.getModificationLocalizationParameters();
        if (modificationScoringPreferences.getAlignNonConfidentModifications()) {
            waitingHandler.appendReport("Resolving peptide inference issues.", true, true);
            peptideInference.peptideInference(identification, identificationParameters, sequenceProvider, modificationFactory, waitingHandler);
            waitingHandler.increasePrimaryProgressCounter();
            if (waitingHandler.isRunCanceled()) {
                return;
            }
        }
        identification.getObjectsDB().commit();
        System.gc();
    }
    String reportTxt = "Saving probabilities";
    String waitingTitle = "Saving Probabilities.";
    ;
    switch(projectType) {
        case psm:
            reportTxt += ".";
            break;
        case peptide:
            reportTxt += ", building peptides.";
            waitingTitle += " Building Peptides.";
            break;
        default:
            reportTxt += ", building peptides and proteins.";
            waitingTitle += " Building Peptides and Proteins.";
    }
    waitingHandler.appendReport(reportTxt, true, true);
    waitingHandler.setWaitingText(waitingTitle + " Please Wait...");
    attachSpectrumProbabilitiesAndBuildPeptidesAndProteins(sequenceProvider, identificationParameters.getSequenceMatchingParameters(), projectType, fastaParameters, waitingHandler);
    waitingHandler.increasePrimaryProgressCounter();
    if (waitingHandler.isRunCanceled()) {
        return;
    }
    identification.getObjectsDB().commit();
    System.gc();
    if (projectType == ProjectType.peptide || projectType == ProjectType.protein) {
        waitingHandler.appendReport("Generating peptide map.", true, true);
        matchesValidator.fillPeptideMaps(identification, metrics, waitingHandler, identificationParameters, sequenceProvider, spectrumProvider);
        if (waitingHandler.isRunCanceled()) {
            return;
        }
        identification.getObjectsDB().commit();
        System.gc();
        waitingHandler.appendReport("Computing peptide probabilities.", true, true);
        matchesValidator.getPeptideMap().estimateProbabilities(waitingHandler);
        if (waitingHandler.isRunCanceled()) {
            return;
        }
        identification.getObjectsDB().commit();
        System.gc();
        waitingHandler.appendReport("Saving peptide probabilities.", true, true);
        matchesValidator.attachPeptideProbabilities(identification, fastaParameters, waitingHandler);
        waitingHandler.increasePrimaryProgressCounter();
        if (waitingHandler.isRunCanceled()) {
            return;
        }
        identification.getObjectsDB().commit();
        System.gc();
        if (projectType == ProjectType.protein) {
            if (identificationParameters.getProteinInferenceParameters().getSimplifyGroups()) {
                waitingHandler.appendReport("Simplifying protein groups.", true, true);
                GroupSimplification groupSimplification = new GroupSimplification();
                groupSimplification.removeRedundantGroups(identification, identificationParameters, sequenceProvider, proteinDetailsProvider, waitingHandler);
                waitingHandler.increasePrimaryProgressCounter();
                if (waitingHandler.isRunCanceled()) {
                    return;
                }
            }
            identification.getObjectsDB().commit();
            System.gc();
            ProteinInference proteinInference = new ProteinInference();
            waitingHandler.appendReport("Mapping shared peptides.", true, true);
            proteinInference.distributeSharedPeptides(identification, waitingHandler);
            waitingHandler.increasePrimaryProgressCounter();
            if (waitingHandler.isRunCanceled()) {
                return;
            }
            identification.getObjectsDB().commit();
            System.gc();
            waitingHandler.appendReport("Generating protein map.", true, true);
            matchesValidator.fillProteinMap(identification, spectrumProvider, waitingHandler);
            waitingHandler.increasePrimaryProgressCounter();
            if (waitingHandler.isRunCanceled()) {
                return;
            }
            identification.getObjectsDB().commit();
            System.gc();
            waitingHandler.appendReport("Selecting leading proteins, inferring peptide and protein inference status.", true, true);
            proteinInference.inferPiStatus(identification, metrics, matchesValidator.getProteinMap(), identificationParameters, sequenceProvider, proteinDetailsProvider, waitingHandler);
            waitingHandler.increasePrimaryProgressCounter();
            if (waitingHandler.isRunCanceled()) {
                return;
            }
            identification.getObjectsDB().commit();
            System.gc();
            waitingHandler.appendReport("Computing protein probabilities.", true, true);
            matchesValidator.getProteinMap().estimateProbabilities(waitingHandler);
            if (waitingHandler.isRunCanceled()) {
                return;
            }
            identification.getObjectsDB().commit();
            System.gc();
            waitingHandler.appendReport("Saving protein probabilities.", true, true);
            matchesValidator.attachProteinProbabilities(identification, sequenceProvider, fastaParameters, metrics, waitingHandler, identificationParameters.getFractionParameters());
            waitingHandler.increasePrimaryProgressCounter();
            if (waitingHandler.isRunCanceled()) {
                return;
            }
            identification.getObjectsDB().commit();
            System.gc();
        }
    }
    if (fastaParameters.isTargetDecoy()) {
        IdMatchValidationParameters idMatchValidationParameters = identificationParameters.getIdValidationParameters();
        if (idMatchValidationParameters.getDefaultPsmFDR() == 1 && idMatchValidationParameters.getDefaultPeptideFDR() == 1 && idMatchValidationParameters.getDefaultProteinFDR() == 1) {
            waitingHandler.appendReport("Validating identifications at 1% FDR, quality control of matches.", true, true);
        } else {
            waitingHandler.appendReport("Validating identifications, quality control of matches.", true, true);
        }
    } else {
        waitingHandler.appendReport("Quality control of matches.", true, true);
    }
    matchesValidator.validateIdentifications(identification, metrics, inputMap, waitingHandler, exceptionHandler, identificationFeaturesGenerator, sequenceProvider, proteinDetailsProvider, spectrumProvider, geneMaps, identificationParameters, projectType, processingParameters);
    waitingHandler.increasePrimaryProgressCounter();
    if (waitingHandler.isRunCanceled()) {
        return;
    }
    identification.getObjectsDB().commit();
    System.gc();
    if (projectType == ProjectType.peptide || projectType == ProjectType.protein) {
        waitingHandler.appendReport("Scoring PTMs in peptides.", true, true);
        modificationLocalizationScorer.scorePeptidePtms(identification, modificationFactory, waitingHandler, identificationParameters);
        waitingHandler.increasePrimaryProgressCounter();
        if (waitingHandler.isRunCanceled()) {
            return;
        }
        identification.getObjectsDB().commit();
        System.gc();
        if (projectType == ProjectType.protein) {
            waitingHandler.appendReport("Estimating spectrum counting scaling values.", true, true);
            ScalingFactorsEstimators scalingFactors = new ScalingFactorsEstimators(spectrumCountingParameters);
            scalingFactors.estimateScalingFactors(identification, metrics, sequenceProvider, identificationFeaturesGenerator, waitingHandler, exceptionHandler, processingParameters);
            waitingHandler.increasePrimaryProgressCounter();
            if (waitingHandler.isRunCanceled()) {
                return;
            }
            identification.getObjectsDB().commit();
            System.gc();
            waitingHandler.appendReport("Scoring PTMs in proteins, gathering summary metrics.", true, true);
            ProteinProcessor proteinProcessor = new ProteinProcessor(identification, identificationParameters, identificationFeaturesGenerator, sequenceProvider);
            proteinProcessor.processProteins(modificationLocalizationScorer, metrics, modificationFactory, waitingHandler, exceptionHandler, processingParameters);
            waitingHandler.increasePrimaryProgressCounter();
            if (waitingHandler.isRunCanceled()) {
                return;
            }
            identification.getObjectsDB().commit();
            System.gc();
        }
    }
    projectCreationDuration.end();
    String report = "Identification processing completed (" + projectCreationDuration.toString() + ").";
    waitingHandler.appendReport(report, true, true);
    waitingHandler.appendReportEndLine();
    waitingHandler.appendReportEndLine();
    identification.addUrParam(new PSMaps(inputMap, matchesValidator.getPsmMap(), matchesValidator.getPeptideMap(), matchesValidator.getProteinMap()));
    if (setWaitingHandlerFinshedWhenDone) {
        waitingHandler.setRunFinished();
    }
}

Example 4 with FastaSummary

use of com.compomics.util.experiment.io.biology.protein.FastaSummary in project peptide-shaker by compomics.

the class CLIExportMethods method exportMzId.

/**
 * Exports the project in the mzIdentML format.
 *
 * @param mzidCLIInputBean the user input
 * @param psbdParent a psbd file parent allowing accessing the information it
 * contains
 * @param waitingHandler a waiting handler allowing display of progress and
 * interruption of the export
 *
 * @throws IOException exception thrown whenever an IO exception occurred
 * while reading or writing to a file
 */
public static void exportMzId(MzidCLIInputBean mzidCLIInputBean, PsdbParent psbdParent, WaitingHandler waitingHandler) throws IOException {
    ProjectDetails projectDetails = psbdParent.getProjectDetails();
    projectDetails.setContactFirstName(mzidCLIInputBean.getContactFirstName());
    projectDetails.setContactLastName(mzidCLIInputBean.getContactLastName());
    projectDetails.setContactEmail(mzidCLIInputBean.getContactEmail());
    projectDetails.setContactAddress(mzidCLIInputBean.getContactAddress());
    projectDetails.setContactUrl(mzidCLIInputBean.getContactUrl());
    projectDetails.setOrganizationName(mzidCLIInputBean.getOrganizationName());
    projectDetails.setOrganizationEmail(mzidCLIInputBean.getOrganizationMail());
    projectDetails.setOrganizationAddress(mzidCLIInputBean.getOrganizationAddress());
    projectDetails.setOrganizationUrl(mzidCLIInputBean.getOrganizationUrl());
    projectDetails.setIncludeProteinSequences(mzidCLIInputBean.getIncludeProteinSequences());
    projectDetails.setPrideOutputFolder(mzidCLIInputBean.getOutputFile().getAbsolutePath());
    IdentificationParameters identificationParameters = psbdParent.getIdentificationParameters();
    FastaSummary fastaSummary = FastaSummary.getSummary(projectDetails.getFastaFile(), identificationParameters.getFastaParameters(), waitingHandler);
    MzIdentMLExport mzIdentMLExport = new MzIdentMLExport(PeptideShaker.getVersion(), psbdParent.getIdentification(), psbdParent.getProjectDetails(), identificationParameters, psbdParent.getSequenceProvider(), psbdParent.getProteinDetailsProvider(), psbdParent.getSpectrumProvider(), ModificationFactory.getInstance(), fastaSummary, psbdParent.getIdentificationFeaturesGenerator(), mzidCLIInputBean.getOutputFile(), mzidCLIInputBean.getIncludeProteinSequences(), waitingHandler, mzidCLIInputBean.isGzip());
    mzIdentMLExport.createMzIdentMLFile(mzidCLIInputBean.getMzIdentMLVersion());
}

Example 5 with FastaSummary

use of com.compomics.util.experiment.io.biology.protein.FastaSummary in project peptide-shaker by compomics.

the class PeptideShakerCLI method createProject.

/**
 * Creates the PeptideShaker project based on the identification files
 * provided in the command line input
 *
 * @throws java.io.IOException exception thrown if an error occurs while
 * reading or writing a file
 * @throws java.lang.InterruptedException exception thrown if a thread is
 * interrupted
 * @throws java.util.concurrent.TimeoutException exception thrown if a
 * process times out
 */
public void createProject() throws IOException, InterruptedException, TimeoutException {
    // define new project reference
    projectParameters = new ProjectParameters(cliInputBean.getExperimentID());
    // set the project details
    projectDetails = new ProjectDetails();
    projectDetails.setCreationDate(new Date());
    projectDetails.setPeptideShakerVersion(new Properties().getVersion());
    // set up spectrum provider
    msFileHandler = new MsFileHandler();
    // get the input files
    ArrayList<File> identificationFilesInput = cliInputBean.getIdFiles();
    ArrayList<File> dataFolders = new ArrayList<>();
    ArrayList<File> spectrumFiles = cliInputBean.getSpectrumFiles();
    File fastaFile = null;
    // Extract data from zip files, try to find the search parameter and spectrum files
    ArrayList<File> identificationFiles = new ArrayList<>();
    IdentificationParameters tempIdentificationParameters = null;
    for (File inputFile : identificationFilesInput) {
        File parentFile = inputFile.getParentFile();
        if (!dataFolders.contains(parentFile)) {
            dataFolders.add(parentFile);
        }
        File dataFolder = new File(parentFile, "mgf");
        if (dataFolder.exists() && !dataFolders.contains(dataFolder)) {
            dataFolders.add(dataFolder);
        }
        dataFolder = new File(parentFile, "mzml");
        if (dataFolder.exists() && !dataFolders.contains(dataFolder)) {
            dataFolders.add(dataFolder);
        }
        dataFolder = new File(parentFile, "cms");
        if (dataFolder.exists() && !dataFolders.contains(dataFolder)) {
            dataFolders.add(dataFolder);
        }
        dataFolder = new File(parentFile, "fasta");
        if (dataFolder.exists() && !dataFolders.contains(dataFolder)) {
            dataFolders.add(dataFolder);
        }
        dataFolder = new File(parentFile, PeptideShaker.DATA_DIRECTORY);
        if (dataFolder.exists() && !dataFolders.contains(dataFolder)) {
            dataFolders.add(dataFolder);
        }
        String fileName = inputFile.getName();
        if (fileName.toLowerCase().endsWith(".zip")) {
            waitingHandler.appendReport("Unzipping " + fileName + ".", true, true);
            String newName = PsZipUtils.getTempFolderName(fileName);
            String parentFolder = PsZipUtils.getUnzipParentFolder();
            if (parentFolder == null) {
                parentFolder = parentFile.getAbsolutePath();
            }
            File parentFolderFile = new File(parentFolder, PsZipUtils.getUnzipSubFolder());
            File destinationFolder = new File(parentFolderFile, newName);
            destinationFolder.mkdir();
            TempFilesManager.registerTempFolder(parentFolderFile);
            ZipUtils.unzip(inputFile, destinationFolder, waitingHandler);
            if (waitingHandler instanceof WaitingHandlerCLIImpl) {
                waitingHandler.appendReportEndLine();
            }
            dataFolder = new File(destinationFolder, PeptideShaker.DATA_DIRECTORY);
            if (dataFolder.exists() && !dataFolders.contains(dataFolder)) {
                dataFolders.add(dataFolder);
            }
            dataFolder = new File(destinationFolder, ".mgf");
            if (dataFolder.exists() && !dataFolders.contains(dataFolder)) {
                dataFolders.add(dataFolder);
            }
            dataFolder = new File(parentFile, "mzml");
            if (dataFolder.exists() && !dataFolders.contains(dataFolder)) {
                dataFolders.add(dataFolder);
            }
            dataFolder = new File(destinationFolder, ".cms");
            if (dataFolder.exists() && !dataFolders.contains(dataFolder)) {
                dataFolders.add(dataFolder);
            }
            dataFolder = new File(destinationFolder, ".fasta");
            if (dataFolder.exists() && !dataFolders.contains(dataFolder)) {
                dataFolders.add(dataFolder);
            }
            for (File unzippedFile : destinationFolder.listFiles()) {
                String nameLowerCase = unzippedFile.getName().toLowerCase();
                if (nameLowerCase.endsWith(".omx") || nameLowerCase.endsWith(".t.xml") || nameLowerCase.endsWith(".pep.xml") || nameLowerCase.endsWith(".dat") || nameLowerCase.endsWith(".mzid") || nameLowerCase.endsWith(".ms-amanda.csv") || nameLowerCase.endsWith(".res") || nameLowerCase.endsWith(".tide-search.target.txt") || nameLowerCase.endsWith(".tags") || nameLowerCase.endsWith(".pnovo.txt") || nameLowerCase.endsWith(".novor.csv") || nameLowerCase.endsWith(".psm") || nameLowerCase.endsWith(".omx.gz") || nameLowerCase.endsWith(".t.xml.gz") || nameLowerCase.endsWith(".pep.xml.gz") || nameLowerCase.endsWith(".mzid.gz") || nameLowerCase.endsWith(".ms-amanda.csv.gz") || nameLowerCase.endsWith(".res.gz") || nameLowerCase.endsWith(".tide-search.target.txt.gz") || nameLowerCase.endsWith(".tags.gz") || nameLowerCase.endsWith(".pnovo.txt.gz") || nameLowerCase.endsWith(".novor.csv.gz") || nameLowerCase.endsWith(".psm.gz")) {
                    identificationFiles.add(unzippedFile);
                } else if (nameLowerCase.endsWith(".par")) {
                    try {
                        tempIdentificationParameters = IdentificationParameters.getIdentificationParameters(unzippedFile);
                        ValidationQcParameters validationQCParameters = tempIdentificationParameters.getIdValidationParameters().getValidationQCParameters();
                        if (validationQCParameters == null || validationQCParameters.getPsmFilters() == null || validationQCParameters.getPeptideFilters() == null || validationQCParameters.getProteinFilters() == null || validationQCParameters.getPsmFilters().isEmpty() && validationQCParameters.getPeptideFilters().isEmpty() && validationQCParameters.getProteinFilters().isEmpty()) {
                            MatchesValidator.setDefaultMatchesQCFilters(validationQCParameters);
                        }
                    } catch (Exception e) {
                        e.printStackTrace();
                        waitingHandler.appendReport("An error occurred while parsing the parameters file " + unzippedFile.getName() + ". " + getLogFileMessage(), true, true);
                        waitingHandler.setRunCanceled();
                    }
                }
            }
        } else {
            identificationFiles.add(inputFile);
        }
    }
    // list the spectrum files found
    HashSet<String> dataFileNamesRequired = new HashSet<>();
    for (File spectrumFile : spectrumFiles) {
        dataFileNamesRequired.add(IoUtil.getFileName(spectrumFile));
    }
    for (File dataFolder : dataFolders) {
        for (File file : dataFolder.listFiles()) {
            String name = file.getName();
            if (name.endsWith(".mgf") || name.endsWith(".mgf.gz") || name.endsWith(".mzml") || name.endsWith(".mzml.gz") || name.endsWith(".cms")) {
                if (!dataFileNamesRequired.contains(name)) {
                    spectrumFiles.add(file);
                    dataFileNamesRequired.add(name);
                }
            } else if (name.endsWith(".fasta")) {
                if (!dataFileNamesRequired.contains(name)) {
                    fastaFile = file;
                    dataFileNamesRequired.add(name);
                }
            }
        }
    }
    // Load the spectrum files
    for (File spectrumFile : spectrumFiles) {
        File folder = CmsFolder.getParentFolder() == null ? spectrumFile.getParentFile() : new File(CmsFolder.getParentFolder());
        msFileHandler.register(spectrumFile, folder, waitingHandler);
    }
    // If there is a specific fasta file chosen, it is used insted of the one included in the searchgui zip
    if (cliInputBean.getFastaFile() != null) {
        fastaFile = cliInputBean.getFastaFile();
    }
    // get the identification parameters
    IdentificationParametersInputBean identificationParametersInputBean = cliInputBean.getIdentificationParametersInputBean();
    if (tempIdentificationParameters != null && identificationParametersInputBean.getInputFile() == null) {
        identificationParametersInputBean.setIdentificationParameters(tempIdentificationParameters);
        identificationParametersInputBean.updateIdentificationParameters();
    }
    identificationParameters = identificationParametersInputBean.getIdentificationParameters();
    ValidationQcParameters validationQCParameters = identificationParameters.getIdValidationParameters().getValidationQCParameters();
    if (validationQCParameters == null || validationQCParameters.getPsmFilters() == null || validationQCParameters.getPeptideFilters() == null || validationQCParameters.getProteinFilters() == null || validationQCParameters.getPsmFilters().isEmpty() && validationQCParameters.getPeptideFilters().isEmpty() && validationQCParameters.getProteinFilters().isEmpty()) {
        MatchesValidator.setDefaultMatchesQCFilters(validationQCParameters);
    }
    if (identificationParameters == null) {
        waitingHandler.appendReport("Identification parameters not found!", true, true);
        waitingHandler.setRunCanceled();
    }
    SearchParameters searchParameters = identificationParameters.getSearchParameters();
    String error = PeptideShaker.loadModifications(searchParameters);
    if (error != null) {
        System.out.println(error);
    }
    // try to locate the fasta file
    if (fastaFile == null) {
        waitingHandler.appendReport("FASTA file not set (or not in zip file)!", true, true);
        waitingHandler.setRunCanceled();
    } else if (!fastaFile.exists()) {
        boolean found = false;
        // look in the database folder
        try {
            File tempDbFolder = utilitiesUserParameters.getDbFolder();
            File newFile = new File(tempDbFolder, fastaFile.getName());
            if (newFile.exists()) {
                fastaFile = newFile;
                projectDetails.setFastaFile(fastaFile);
                found = true;
            }
        } catch (Exception e) {
            e.printStackTrace();
        }
        if (!found) {
            // look in the data folders
            for (File dataFolder : dataFolders) {
                File newFile = new File(dataFolder, fastaFile.getName());
                if (newFile.exists()) {
                    fastaFile = newFile;
                    projectDetails.setFastaFile(fastaFile);
                    found = true;
                    break;
                }
            }
            if (!found) {
                waitingHandler.appendReport("FASTA file \'" + fastaFile.getName() + "\' not found.", true, true);
                waitingHandler.setRunCanceled();
            }
        }
    } else {
        projectDetails.setFastaFile(fastaFile);
    }
    // get the summary information for the FASTA file
    try {
        // get the FASTA summary
        FastaSummary fastaSummary = loadFastaFile(waitingHandler);
        // set the background species
        identificationParameters.getGeneParameters().setBackgroundSpeciesFromFastaSummary(fastaSummary);
    } catch (IOException e) {
        e.printStackTrace();
        waitingHandler.appendReport("An error occurred while parsing the FASTA file.", true, true);
        waitingHandler.setRunCanceled();
    }
    // set the processing settings
    ProcessingParameters processingParameters = new ProcessingParameters();
    Integer nThreads = cliInputBean.getnThreads();
    if (nThreads != null) {
        processingParameters.setnThreads(nThreads);
    }
    // set the spectrum counting prefrences
    spectrumCountingParameters = new SpectrumCountingParameters();
    // set the project type
    projectType = cliInputBean.getProjectType();
    // check the project reference
    for (String forbiddenChar : Util.FORBIDDEN_CHARACTERS) {
        if (cliInputBean.getExperimentID().contains(forbiddenChar)) {
            waitingHandler.appendReport("The project name cannot not contain " + forbiddenChar + ".", true, true);
            waitingHandler.setRunCanceled();
        }
    }
    // incrementing the counter for a new PeptideShaker start run via CLI
    if (utilitiesUserParameters.isAutoUpdate()) {
        Util.sendGAUpdate("UA-36198780-1", "startrun-cl", "peptide-shaker-" + PeptideShaker.getVersion());
    }
    // create a shaker which will perform the analysis
    PeptideShaker peptideShaker = new PeptideShaker(projectParameters);
    // import the files
    int outcome = peptideShaker.importFiles(waitingHandler, identificationFiles, msFileHandler, identificationParameters, projectDetails, processingParameters, exceptionHandler);
    if (outcome == 0) {
        peptideShaker.createProject(identificationParameters, processingParameters, spectrumCountingParameters, msFileHandler, projectDetails, projectType, waitingHandler, false, exceptionHandler);
    }
    if (!waitingHandler.isRunCanceled()) {
        // identification as created by PeptideShaker
        identification = peptideShaker.getIdentification();
        // metrics saved while processing the data
        metrics = peptideShaker.getMetrics();
        // fene maps
        geneMaps = peptideShaker.getGeneMaps();
        // the identification feature generator
        identificationFeaturesGenerator = peptideShaker.getIdentificationFeaturesGenerator();
        // the sequence provider
        sequenceProvider = peptideShaker.getSequenceProvider();
        // the protein details provider
        proteinDetailsProvider = peptideShaker.getProteinDetailsProvider();
        if (waitingHandler instanceof WaitingDialog) {
            projectDetails.setReport(((WaitingDialog) waitingHandler).getReport(null));
            ((WaitingDialog) waitingHandler).setRunNotFinished();
            ((WaitingDialog) waitingHandler).setCloseDialogWhenImportCompletes(true, false);
        }
    } else {
        if (waitingHandler instanceof WaitingDialog) {
            saveReport();
        }
        TempFilesManager.deleteTempFolders();
        waitingHandler.setWaitingText("PeptideShaker Processing Canceled.");
        System.out.println("<CompomicsError>PeptideShaker processing canceled. " + getLogFileMessage() + "</CompomicsError>");
    }
}