public MsToLcmsFeatures(IScanSummaryProvider provider, LcmsFeatureFindingOptions options = null) { if (provider == null) { throw new ArgumentNullException(); } Comparison <MSFeatureLight> mzSort = (x, y) => x.Mz.CompareTo(y.Mz); Comparison <UMCLight> monoSort = (x, y) => x.MassMonoisotopic.CompareTo(y.MassMonoisotopic); Func <MSFeatureLight, MSFeatureLight, double> mzDiff = (x, y) => FeatureLight.ComputeMassPPMDifference(x.Mz, y.Mz); Func <UMCLight, UMCLight, double> monoDiff = (x, y) => FeatureLight.ComputeMassPPMDifference(x.MassMonoisotopic, y.MassMonoisotopic); this.provider = provider; this.options = options ?? new LcmsFeatureFindingOptions(); // Set clusterers if (this.options.FirstPassClusterer == MsFeatureClusteringAlgorithmType.BinarySearchTree) { this.firstPassClusterer = new MsFeatureTreeClusterer <MSFeatureLight, UMCLight>( mzSort, mzDiff, MassComparison.Mz, this.options.InstrumentTolerances.Mass); } else { this.firstPassClusterer = ClusterFactory.Create(this.options.FirstPassClusterer); } if (this.options.SecondPassClusterer == GenericClusteringAlgorithmType.BinarySearchTree) { this.secondPassClusterer = new MsFeatureTreeClusterer <UMCLight, UMCLight>( monoSort, monoDiff, MassComparison.Monoisotopic, this.options.InstrumentTolerances.Mass); } else { var clusterFactory = new GenericClusterFactory <UMCLight, UMCLight>(); this.secondPassClusterer = clusterFactory.Create(this.options.SecondPassClusterer); } }
public void TestClustering( string directory, string outputPath, FeatureAlignmentType alignmentType, LcmsFeatureClusteringAlgorithmType clusterType) { var matchPath = string.Format("{0}.txt", outputPath); var errorPath = string.Format("{0}-errors.txt", outputPath); // Loads the supported MultiAlign types var supportedTypes = DatasetLoader.SupportedFileTypes; var extensions = new List <string>(); supportedTypes.ForEach(x => extensions.Add("*" + x.Extension)); // Find our datasets var datasetLoader = new DatasetLoader(); var datasets = datasetLoader.GetValidDatasets(directory, extensions, SearchOption.TopDirectoryOnly); // Setup our alignment options var alignmentOptions = new AlignmentOptions(); var spectralOptions = new SpectralOptions { ComparerType = SpectralComparison.CosineDotProduct, Fdr = .01, IdScore = 1e-09, MzBinSize = .5, MzTolerance = .5, NetTolerance = .1, RequiredPeakCount = 32, SimilarityCutoff = .75, TopIonPercent = .8 }; // Options setup var instrumentOptions = InstrumentPresetFactory.Create(InstrumentPresets.LtqOrbitrap); var featureTolerances = new FeatureTolerances { Mass = instrumentOptions.Mass + 6, Net = instrumentOptions.NetTolerance, DriftTime = instrumentOptions.DriftTimeTolerance }; var featureFindingOptions = new LcmsFeatureFindingOptions(featureTolerances) { MaximumNetRange = .002, MaximumScanRange = 50 }; // Create our algorithms var finder = FeatureFinderFactory.CreateFeatureFinder(FeatureFinderType.TreeBased); var aligner = FeatureAlignerFactory.CreateDatasetAligner(alignmentType, alignmentOptions.LCMSWarpOptions, spectralOptions); var clusterer = ClusterFactory.Create(clusterType); clusterer.Parameters = new FeatureClusterParameters <UMCLight> { Tolerances = featureTolerances }; RegisterProgressNotifier(aligner); RegisterProgressNotifier(finder); RegisterProgressNotifier(clusterer); var lcmsFilters = new LcmsFeatureFilteringOptions { FeatureLengthRangeScans = new FilterRange(50, 300) }; var msFilterOptions = new MsFeatureFilteringOptions { MinimumIntensity = 5000, ChargeRange = new FilterRange(1, 6), ShouldUseChargeFilter = true, ShouldUseDeisotopingFilter = true, ShouldUseIntensityFilter = true }; for (var i = 0; i < 1; i++) { var aligneeDatasets = datasets.Where((t, j) => j != i).ToList(); PerformMultiAlignAnalysis(datasets[0], aligneeDatasets, featureFindingOptions, msFilterOptions, lcmsFilters, spectralOptions, finder, aligner, clusterer, matchPath, errorPath); } }
public void GenerateClusterAlignmentStatistics(string relativeDatabasePath, string relativeName, string name, FeatureAlignmentType alignmentType, LcmsFeatureClusteringAlgorithmType clusterType) { var databasePath = GetPath(relativeDatabasePath); var outputPath = GetOutputPath(relativeName); if (!Directory.Exists(outputPath)) { Directory.CreateDirectory(outputPath); } // Connect to the NHibernate database var providers = DataAccessFactory.CreateDataAccessProviders(databasePath, false); // Setup our alignment options var alignmentOptions = new AlignmentOptions(); var spectralOptions = new SpectralOptions { ComparerType = SpectralComparison.CosineDotProduct, Fdr = .01, IdScore = 1e-09, MzBinSize = .5, MzTolerance = .5, NetTolerance = .1, RequiredPeakCount = 32, SimilarityCutoff = .75, TopIonPercent = .8 }; // Options setup var instrumentOptions = InstrumentPresetFactory.Create(InstrumentPresets.LtqOrbitrap); var featureTolerances = new FeatureTolerances { Mass = instrumentOptions.Mass + 6, Net = instrumentOptions.NetTolerance, DriftTime = instrumentOptions.DriftTimeTolerance }; UpdateStatus("Retrieving all datasets for test."); var datasets = providers.DatasetCache.FindAll(); // Create our algorithms var aligner = FeatureAlignerFactory.CreateDatasetAligner(alignmentType, alignmentOptions.LCMSWarpOptions, spectralOptions); var clusterer = ClusterFactory.Create(clusterType); clusterer.Parameters = new FeatureClusterParameters <UMCLight> { Tolerances = featureTolerances }; RegisterProgressNotifier(aligner); RegisterProgressNotifier(clusterer); for (var i = 0; i < datasets.Count - 1; i++) { var matchPath = string.Format("{0}-{1}-matches.txt", name, i); var errorPath = string.Format("{0}-{1}-errors.txt", name, i); matchPath = Path.Combine(outputPath, matchPath); errorPath = Path.Combine(outputPath, errorPath); var aligneeDataset = datasets[i + 1]; var baselineDataset = datasets[i]; // Load the baseline reference set using (var rawProviderX = new InformedProteomicsReader()) { rawProviderX.AddDataFile(baselineDataset.RawFile.Path, 0); // Load the baseline reference set using (var rawProviderY = new InformedProteomicsReader()) { rawProviderY.AddDataFile(aligneeDataset.RawFile.Path, 0); var baselineFeatures = RetrieveFeatures(baselineDataset.DatasetId, providers); var aligneeFeatures = RetrieveFeatures(aligneeDataset.DatasetId, providers); var providerX = new CachedFeatureSpectraProvider(rawProviderX, baselineFeatures); var providerY = new CachedFeatureSpectraProvider(rawProviderY, aligneeFeatures); AlignDatasets(baselineFeatures, aligneeFeatures, providerX, providerY, aligner, clusterer, matchPath, errorPath); } } } }