public void LoadFactorsFromDMS(ObservableCollection<DatasetInformation> datasets,
FeatureDataAccessProviders providers)
{
if (datasets.Count < 1)
{
return;
}
var sink = new MultiAlignFactorSink(datasets,
providers.DatasetCache,
providers.FactorCache,
providers.FactorAssignmentCache);
foreach (var info in datasets)
{
var query = m_sqlQuery + string.Format(" = '{0}'", info.DatasetName);
var reader = new MSSQLReader();
reader.Server = Server;
reader.Database = Database;
reader.SQLText = query;
var pipeline = ProcessingPipeline.Assemble("PlainFactors", reader, sink);
pipeline.RunRoot(null);
}
sink.CommitChanges();
}
public void CalculateClusterErrorHistograms(FeatureDataAccessProviders providers, List<double> massErrorPpm, List<double> netError, List<double> counts, FeatureTolerances tolerances)
{
List<UMCLight> featuresA = providers.FeatureCache.FindByDatasetId(0);
List<UMCLight> featuresB = providers.FeatureCache.FindByDatasetId(1);
featuresA.Sort(
delegate (UMCLight x, UMCLight y)
{
return x.MassMonoisotopic.CompareTo(y.MassMonoisotopic);
}
);
featuresB.Sort(
delegate (UMCLight x, UMCLight y)
{
return x.MassMonoisotopic.CompareTo(y.MassMonoisotopic);
}
);
int i = 0;
foreach(UMCLight featureA in featuresA)
{
double count = 0;
int j = i + 1;
foreach (UMCLight featureB in featuresB)
{
double ppmDiff = Feature.ComputeMassPPMDifference(featureB.MassMonoisotopicAligned, featureA.MassMonoisotopicAligned);
if (Math.Abs(ppmDiff) > tolerances.Mass)
{
continue;
}
double netDiff = featureA.RetentionTime - featureB.RetentionTime;
if (Math.Abs(netDiff) > tolerances.RetentionTime)
{
continue;
}
massErrorPpm.Add(ppmDiff);
netError.Add(netDiff);
}
i = j;
counts.Add(count);
}
}
public static FeatureDataAccessProviders CreateDataAccessProviders(string path, bool deleteIfExists)
{
if (deleteIfExists)
{
var exists = File.Exists(path);
if (exists)
{
File.Delete(path);
}
}
try
{
NHibernateUtil.ConnectToDatabase(path, true);
}
catch (Exception ex)
{
throw ex;
}
IUmcDAO featureCache = new UmcDAOHibernate();
IUmcClusterDAO clusterCache = new UmcClusterDAOHibernate();
IMSFeatureDAO msFeatureCache = new MSFeatureDAOHibernate();
IMSnFeatureDAO msnFeatureCache = new MSnFeatureDAOHibernate();
IMsnFeatureToMSFeatureDAO msnMSCache = new MSnFeatureToMSFeatureDAOHibernate();
IDatasetDAO datasetCache = new DatasetDAOHibernate();
IMassTagDAO massTagCache = new MassTagDAOHibernate();
IMassTagMatchDAO massTagMatchCache = new MassTagMatchDAO();
ISequenceToMsnFeatureDAO sequenceMap = new SequenceToMsnDAOHibernate();
var providers =
new FeatureDataAccessProviders(featureCache,
clusterCache,
msFeatureCache,
msnFeatureCache,
msnMSCache,
datasetCache,
massTagMatchCache,
massTagCache,
new FactorDAOHibernate(),
new DatasetToFactorDAOHibernate(),
new MSMSClusterMapClusterDAOHibernate(),
new DatabaseSearchSequenceDAOHiberate(),
sequenceMap);
return providers;
}
public void LoadFactorsFromFile(string path, ObservableCollection<DatasetInformation> datasets,
FeatureDataAccessProviders providers)
{
if (datasets.Count < 1)
{
return;
}
var sink = new MultiAlignFactorSink(datasets,
providers.DatasetCache,
providers.FactorCache,
providers.FactorAssignmentCache);
var reader = new DelimitedFileReader();
reader.Delimiter = "\t";
reader.FilePath = path;
var pipeline = ProcessingPipeline.Assemble("PlainFactors", reader, sink);
pipeline.RunRoot(null);
sink.CommitChanges();
}
private IEnumerable<UMCLight> RetrieveFeatures(int datasetId, FeatureDataAccessProviders providers)
{
var features = providers.FeatureCache.FindByDatasetId(datasetId);
var spectra = providers.MSnFeatureCache.FindByDatasetId(datasetId);
if (spectra == null)
throw new ArgumentNullException(@"There were no MS/MS spectra in the database");
var sequences = providers.DatabaseSequenceCache.FindAll();
var sequenceMaps = providers.SequenceMsnMapCache.FindByDatasetId(datasetId);
var spectraMaps = providers.MSFeatureToMSnFeatureCache.FindByDatasetId(datasetId);
var msFeatures = providers.MSFeatureCache.FindByDatasetId(datasetId);
// Make a one pass through each enumerable list,
// then use the maps to join the data together
var dictFeatures = new Dictionary<int, UMCLight>();
var dictSpectra = new Dictionary<int, MSSpectra>();
var dictPeptide = new Dictionary<int, Peptide>();
var dictMsFeatures = new Dictionary<int, MSFeatureLight>();
foreach (var sequence in sequences)
{
if (sequence.GroupId != datasetId)
continue;
var peptide = new Peptide
{
Sequence = sequence.Sequence,
Id = sequence.Id,
};
dictPeptide.Add(peptide.Id, peptide);
}
msFeatures.ForEach(x => dictMsFeatures.Add(x.Id, x));
features.ForEach(x => dictFeatures.Add(x.Id, x));
spectra.ForEach(x => dictSpectra.Add(x.Id, x));
var count = 0;
// Map the MSMS
foreach (var map in sequenceMaps)
{
MSSpectra spectrum;
Peptide peptide;
var workedSpectra = dictSpectra.TryGetValue(map.MsnFeatureId, out spectrum);
var workedPeptide = dictPeptide.TryGetValue(map.SequenceId, out peptide);
if (workedSpectra && workedPeptide)
{
spectrum.Peptides.Add(peptide);
peptide.Spectrum = spectrum;
count++;
}
}
Console.WriteLine("Mapped {0} peptides to spectra", count);
count = 0;
// Map the spectra....
foreach (var map in spectraMaps)
{
UMCLight feature;
MSSpectra spectrum;
MSFeatureLight msFeature;
var workedFeatures = dictFeatures.TryGetValue(map.LCMSFeatureID, out feature);
var workedSpectra = dictSpectra.TryGetValue(map.MSMSFeatureID, out spectrum);
var workedMsFeature = dictMsFeatures.TryGetValue(map.MSFeatureID, out msFeature);
if (!workedFeatures || !workedSpectra || !workedMsFeature)
continue;
var metaData = new ScanSummary
{
MsLevel = 2,
PrecursorMz = spectrum.PrecursorMz,
Scan = spectrum.Scan
};
spectrum.ScanMetaData = metaData;
msFeature.MSnSpectra.Add(spectrum);
spectrum.ParentFeature = msFeature;
feature.AddChildFeature(msFeature);
msFeature.Umc = feature;
count++;
}
Console.WriteLine("Mapped {0} spectra to parent Features", count);
return features;
}
private FeatureDataAccessProviders SetupDataProviders(bool createNewDatabase)
{
FeatureDataAccessProviders providers;
Logger.PrintMessage("Setting up data providers for caching and storage.");
try
{
var path = AnalysisPathUtils.BuildAnalysisName(m_config.AnalysisPath, m_config.AnalysisName);
providers = SetupDataProviders(path, createNewDatabase);
}
catch (IOException ex)
{
Logger.PrintMessage(ex.Message);
Logger.PrintMessage(ex.StackTrace);
throw;
}
return providers;
}
/// <summary>
/// Loads baseline data for alignment.
/// </summary>
private IList<UMCLight> LoadBaselineData(DatasetInformation baselineInfo,
MsFeatureFilteringOptions msFilterOptions,
LcmsFeatureFindingOptions lcmsFindingOptions,
LcmsFeatureFilteringOptions lcmsFilterOptions,
FeatureDataAccessProviders dataProviders,
MassTagDatabase database,
bool shouldUseMassTagDbAsBaseline)
{
IList<UMCLight> baselineFeatures = null;
UpdateStatus("Loading baseline features.");
if (!shouldUseMassTagDbAsBaseline)
{
if (baselineInfo == null)
{
throw new Exception("The baseline dataset was never set.");
}
var cache = new FeatureLoader
{
Providers = dataProviders
};
RegisterProgressNotifier(cache);
UpdateStatus("Loading baseline features from " + baselineInfo.DatasetName + " for alignment.");
baselineFeatures = cache.LoadDataset(baselineInfo,
msFilterOptions,
lcmsFindingOptions,
lcmsFilterOptions);
cache.CacheFeatures(baselineFeatures);
if (BaselineFeaturesLoaded != null)
{
BaselineFeaturesLoaded(this,
new BaselineFeaturesLoadedEventArgs(baselineInfo, baselineFeatures.ToList()));
}
DeRegisterProgressNotifier(cache);
}
else
{
if (database == null)
throw new NullReferenceException(
"The mass tag database has to have data in it if it's being used for drift time alignment.");
UpdateStatus("Setting baseline features for post drift time alignment from mass tag database.");
var tags = FeatureDataConverters.ConvertToUMC(database.MassTags);
if (BaselineFeaturesLoaded == null)
return tags;
if (tags != null)
BaselineFeaturesLoaded(this, new BaselineFeaturesLoadedEventArgs(null, tags.ToList(), database));
}
return baselineFeatures;
}
/// <summary>
/// Determines if MS/MS should also be discovered.
/// </summary>
/// <param name="cluster"></param>
/// <param name="providers"></param>
/// <param name="getMsMS"></param>
public static void ReconstructUMCCluster(this UMCClusterLight cluster, FeatureDataAccessProviders providers,
bool getMsFeature, bool getMsMs)
{
cluster.Features.Clear();
var features = providers.FeatureCache.FindByClusterID(cluster.Id);
var totalSpectra = 0;
var totalIdentified = 0;
foreach (var feature in features)
{
cluster.AddChildFeature(feature);
if (getMsFeature)
{
feature.ReconstructUMC(providers, getMsMs);
foreach (var msFeature in feature.MsFeatures)
{
totalSpectra += msFeature.MSnSpectra.Count;
foreach (var spectrum in msFeature.MSnSpectra)
{
if (spectrum.Peptides.Count > 0)
totalIdentified++;
}
}
}
}
cluster.IdentifiedSpectraCount = totalIdentified;
cluster.MsMsCount = totalSpectra;
}
/// <summary>
/// Gets a cluster and it's subsequent data structures.
/// </summary>
/// <param name="cluster"></param>
/// <param name="providers"></param>
public static void ReconstructUMCCluster(this UMCClusterLight cluster, FeatureDataAccessProviders providers)
{
cluster.ReconstructUMCCluster(providers, true, true);
}
/// <summary>
/// Retrieves a list of known peptides attributed to this cluster.
/// </summary>
/// <param name="cluster"></param>
/// <param name="providers"></param>
/// <returns></returns>
public static List<MSSpectra> FindSpectra(this UMCClusterLight cluster, FeatureDataAccessProviders providers)
{
var peptides = new List<MSSpectra>();
return peptides;
}
private void ExportData(FeatureDataAccessProviders providers, List<DatasetInformation> datasets)
{
if (m_config.ClusterExporters.Count > 0)
{
var clusters = providers.ClusterCache.FindAll();
if (clusters.Count < 1)
{
Logger.PrintMessage("No clusters present in the database.");
return;
}
var features = providers.FeatureCache.FindAllClustered();
var clusterFeatureMap = new Dictionary<int, UMCClusterLight>();
foreach (var cluster in clusters)
{
clusterFeatureMap[cluster.Id] = cluster;
}
foreach (var umc in features)
{
clusterFeatureMap[umc.ClusterId].AddChildFeature(umc);
}
Logger.PrintMessage("Checking for mass tag matches");
var clusterMatches = providers.MassTagMatches.FindAll();
Logger.PrintMessage("Checking for mass tags");
var massTags = providers.MassTags.FindAll();
var clusterMap = new Dictionary<int, List<ClusterToMassTagMap>>();
if (clusterMatches.Count > 0)
{
foreach (var map in clusterMatches)
{
if (!clusterMap.ContainsKey(map.ClusterId))
{
clusterMap.Add(map.ClusterId, new List<ClusterToMassTagMap>());
}
clusterMap[map.ClusterId].Add(map);
}
}
var tags = new Dictionary<string, MassTagLight>();
if (massTags.Count > 0)
{
foreach (var tag in massTags)
{
var key = tag.ConformationId + "-" + tag.Id;
if (!tags.ContainsKey(key))
{
tags.Add(key, tag);
}
}
}
Logger.PrintMessage("Exporting Data");
foreach (var writer in m_config.ClusterExporters)
{
Logger.PrintMessage("Exporting in " + writer + " format to " + m_config.AnalysisPath);
writer.WriteClusters(clusters,
clusterMap,
datasets,
tags);
}
}
}
/// <summary>
/// Loads factors from file or other.
/// </summary>
private void ConstructFactorInformation(InputAnalysisInfo analysisSetupInformation,
ObservableCollection<DatasetInformation> datasets, FeatureDataAccessProviders providers)
{
var mage = new MAGEFactorAdapter();
if (analysisSetupInformation.FactorFile == null)
{
Logger.PrintMessage("Loading Factor Information from DMS");
mage.LoadFactorsFromDMS(datasets, providers);
}
else
{
Logger.PrintMessage("Loading Factor Information from file: " + analysisSetupInformation.FactorFile);
mage.LoadFactorsFromFile(analysisSetupInformation.FactorFile, datasets, providers);
}
}
/// <summary>
/// Creates the analysis processor and synchronizs the events.
/// </summary>
/// <param name="builder"></param>
/// <param name="providers"></param>
/// <returns></returns>
private MultiAlignAnalysisProcessor ConstructAnalysisProcessor(AlgorithmBuilder builder,
FeatureDataAccessProviders providers)
{
var processor = new MultiAlignAnalysisProcessor();
processor.AnalysisStarted += processor_AnalysisStarted;
processor.AnalysisError += processor_AnalysisError;
processor.FeaturesAligned += processor_FeaturesAligned;
processor.FeaturesLoaded += processor_FeaturesLoaded;
processor.MassTagsLoaded += processor_MassTagsLoaded;
processor.FeaturesClustered += processor_FeaturesClustered;
processor.FeaturesPeakMatched += processor_FeaturesPeakMatched;
processor.AnalysisComplete += processor_AnalysisComplete;
processor.Progress += processor_Progress;
processor.BaselineFeaturesLoaded += processor_BaselineFeaturesLoaded;
m_config.Analysis.DataProviders = providers;
processor.AlgorithmProviders = builder.GetAlgorithmProvider(m_config.Analysis.Options);
return processor;
}
