private async Task AddDatasets(string folderPath)
{
var supportedTypes = DatasetLoader.SupportedFileTypes;
var extensions = new List <string>();
supportedTypes.ForEach(x => extensions.Add("*" + x.Extension));
var datasetLoader = new DatasetLoader();
var datasets = datasetLoader.GetValidDatasets(folderPath, extensions, SearchOption.TopDirectoryOnly);
if (!string.IsNullOrEmpty(datasetLoader.ErrorMessage))
{
MessageBox.Show(datasetLoader.ErrorMessage);
}
if (!this.CheckDatasets(datasets))
{
MessageBox.Show("Datasets are incompatible.");
return;
}
// Add valid datasets.
this.Analysis.MetaData.Datasets.AddRange(datasets);
await this.UpdateDatasets();
}
/// <summary>
/// Adds the folder containing dataset specific elements
/// </summary>
public void AddFolderDelegate()
{
var supportedTypes = DatasetLoader.SupportedFileTypes;
var extensions = new List <string>();
supportedTypes.ForEach(x => extensions.Add("*" + x.Extension));
var option = ShouldSearchSubDirectories;
if (DataFolderPath == null)
{
ApplicationStatusMediator.SetStatus("The directory specified does not exist.");
return;
}
if (!Directory.Exists(DataFolderPath))
{
ApplicationStatusMediator.SetStatus("The directory specified does not exist.");
return;
}
var datasetLoader = new DatasetLoader();
var files = datasetLoader.GetValidDatasets(DataFolderPath,
extensions,
option);
AddDatasets(files);
}
private List <DatasetInformation> GetAndValidateDatasets(IEnumerable <string> filePaths)
{
var datasetLoader = new DatasetLoader();
var datasets = datasetLoader.GetValidDatasets(filePaths);
if (!string.IsNullOrEmpty(datasetLoader.ErrorMessage))
{
MessageBox.Show(datasetLoader.ErrorMessage);
}
return(datasets);
}
/// <summary>
/// Adds datasets from a single file
/// </summary>
private void AddSingleFileDelegate()
{
var fileExists = File.Exists(SingleFilePath);
if (fileExists)
{
var datasetLoader = new DatasetLoader();
AddDatasets(datasetLoader.GetValidDatasets(new List <string> {
SingleFilePath
}));
}
else
{
ApplicationStatusMediator.SetStatus("The input file does not exist.");
}
}
/// <summary>
/// Adds a new dataset to the list.
/// </summary>
/// <param name="inputFiles"></param>
/// <returns>A list of added datasets</returns>
public List <DatasetInformation> AddInputFiles(List <InputFile> inputFiles)
{
var datasetLoader = new DatasetLoader();
var addedSets = datasetLoader.GetValidDatasets(inputFiles, false);
this.Datasets.AddRange(addedSets);
// Reformat their Id's
var id = 0;
foreach (var x in Datasets)
{
x.DatasetId = id++;
}
return(addedSets);
}
private async Task AddDatasets(IEnumerable <string> files)
{
var datasetLoader = new DatasetLoader();
var datasets = datasetLoader.GetValidDatasets(files);
if (!string.IsNullOrEmpty(datasetLoader.ErrorMessage))
{
MessageBox.Show(datasetLoader.ErrorMessage);
}
if (!this.CheckDatasets(datasets))
{
MessageBox.Show("Datasets are incompatible.");
return;
}
// Add valid datasets.
this.Analysis.MetaData.Datasets.AddRange(datasets);
await this.UpdateDatasets();
}
/// <summary>
/// Adds a MultiAlign file
/// </summary>
private void AddInputFileDelegate()
{
var fileExists = File.Exists(InputFilePath);
if (fileExists)
{
// Read input files
try
{
var datasetLoader = new DatasetLoader();
var info = MultiAlignFileInputReader.ReadInputFile(InputFilePath);
AddDatasets(datasetLoader.GetValidDatasets(info.Files, false));
}
catch
{
ApplicationStatusMediator.SetStatus("Could not read the input file. Check the file format.");
}
}
else
{
ApplicationStatusMediator.SetStatus("The input file does not exist.");
}
}
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 TestPeptideBands(string directory,
string matchPath)
{
// 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);
// Options setup
var instrumentOptions = InstrumentPresetFactory.Create(InstrumentPresets.LtqOrbitrap);
var featureTolerances = new FeatureTolerances
{
Mass = instrumentOptions.Mass,
Net = instrumentOptions.NetTolerance,
DriftTime = instrumentOptions.DriftTimeTolerance
};
var msFilterOptions = new MsFeatureFilteringOptions
{
MinimumIntensity = 5000,
ChargeRange = new FilterRange(1, 6),
ShouldUseChargeFilter = true,
ShouldUseDeisotopingFilter = true,
ShouldUseIntensityFilter = true
};
var featureFindingOptions = new LcmsFeatureFindingOptions(featureTolerances)
{
MaximumNetRange = .002,
MaximumScanRange = 50
};
var baselineDataset = datasets[0];
UpdateStatus("Loading baseline features.");
var msFeatures = UmcLoaderFactory.LoadMsFeatureData(baselineDataset.Features.Path);
msFeatures = LcmsFeatureFilters.FilterMsFeatures(msFeatures, msFilterOptions);
var finderFinder = FeatureFinderFactory.CreateFeatureFinder(FeatureFinderType.TreeBased);
var peptideOptions = new SpectralOptions
{
ComparerType = SpectralComparison.CosineDotProduct,
Fdr = .05,
IdScore = 1e-09,
MzBinSize = .5,
MzTolerance = .5,
NetTolerance = .1,
RequiredPeakCount = 32,
SimilarityCutoff = .75,
TopIonPercent = .8
};
var features = new List <MSFeatureLight>();
// Load the baseline reference set
using (var rawProviderX = RawLoaderFactory.CreateFileReader(baselineDataset.RawFile.Path))
{
rawProviderX.AddDataFile(baselineDataset.RawFile.Path, 0);
UpdateStatus("Creating Baseline LCMS Features.");
var baselineFeatures = finderFinder.FindFeatures(msFeatures,
featureFindingOptions,
rawProviderX);
LinkPeptidesToFeatures(baselineDataset.Sequence.Path,
baselineFeatures,
peptideOptions.Fdr,
peptideOptions.IdScore);
baselineFeatures.ForEach(x => features.AddRange(x.MsFeatures));
features = features.Where(x => x.HasMsMs()).ToList();
features = features.OrderBy(x => x.Mz).ToList();
var peptideList = new List <MSFeatureLight>();
foreach (var feature in features)
{
foreach (var spectrum in feature.MSnSpectra)
{
var peptideFound = false;
foreach (var peptide in spectrum.Peptides)
{
peptideList.Add(feature);
peptideFound = true;
break;
}
if (peptideFound)
{
break;
}
}
}
using (var writer = File.CreateText(matchPath))
{
writer.WriteLine("Charge\tpmz\tscan\tNET\t");
foreach (var feature in peptideList)
{
writer.WriteLine("{0}\t{1}\t{2}\t{3}\t", feature.ChargeState, feature.Mz, feature.Scan,
feature.Net);
}
}
}
}
public void CreateFeatureDatabase(string directoryPath, string databasePath)
{
var directory = GetPath(directoryPath);
databasePath = GetPath(databasePath);
// 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);
// 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
};
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
};
var spectralOptions = new SpectralOptions
{
ComparerType = SpectralComparison.CosineDotProduct,
Fdr = .01,
IdScore = 1e-09,
MzBinSize = .5,
MzTolerance = .5,
NetTolerance = .1,
RequiredPeakCount = 32,
SimilarityCutoff = .75,
TopIonPercent = .8
};
var finder = FeatureFinderFactory.CreateFeatureFinder(FeatureFinderType.TreeBased);
NHibernateUtil.CreateDatabase(databasePath);
// Synchronization and IO for serializing all data to the database.
var providers = DataAccessFactory.CreateDataAccessProviders(databasePath, true);
var cache = new FeatureLoader
{
Providers = providers
};
var datasetId = 0;
foreach (var dataset in datasets)
{
dataset.DatasetId = datasetId++;
var features = FindFeatures(dataset,
featureFindingOptions,
msFilterOptions,
lcmsFilters,
spectralOptions,
finder);
cache.CacheFeatures(features);
}
providers.DatasetCache.AddAll(datasets);
}