/// <summary>
/// Aligns the dataset to the data stored in the alignment processor.
/// </summary>
/// <param name="alignmentProcessor">Aligner</param>
/// <param name="features">LC-MS Features to align to the baseline</param>
/// <param name="alignmentOptions">Options</param>
/// <param name="progress"></param>
/// <returns></returns>
private AlignmentData AlignFeatures(LcmsWarpAlignmentProcessor alignmentProcessor,
IEnumerable <UMCLight> features,
LcmsWarpAlignmentOptions alignmentOptions,
IProgress <ProgressData> progress = null)
{
var progData = new ProgressData(progress);
var localProgress = new Progress <ProgressData>(p => progData.Report(p.Percent, p.Status));
var alignmentData = new AlignmentData();
OnStatus("Starting alignment of features.");
// Set minMtdbnet and maxMtdbnet to 0
alignmentData.MinMTDBNET = 0;
alignmentData.MaxMTDBNET = 0;
var umcLights = features as List <UMCLight> ?? features.ToList();
progData.StepRange(5, "Starting alignment of features.");
var filteredFeatures = FilterFeaturesByAbundance(umcLights, alignmentOptions);
// Convert the features, and make a map, so that we can re-adjust the aligned values later.
var map = FeatureDataConverters.MapFeature(umcLights);
progData.StepRange(10, "Setting alignee features.");
// Set features
OnStatus("Setting alignee features.");
alignmentProcessor.SetAligneeDatasetFeatures(filteredFeatures);
progData.StepRange(90, "Performing alignment warping.");
// Find alignment
OnStatus("Performing alignment warping.");
alignmentProcessor.PerformAlignmentToMsFeatures(localProgress);
progData.StepRange(95);
// Extract alignment function
alignmentData.AlignmentFunction = alignmentProcessor.GetAlignmentFunction();
progData.StepRange(100);
// Extract the NET value for every scan
_scanToNETMap = alignmentProcessor.GetScanToNETMapping();
// Correct the features (updates NetAligned and MassMonoisotopicAligned)
OnStatus("Applying alignment function to all features.");
progData.Status = "Applying alignment function to all features.";
umcLights = alignmentProcessor.ApplyNetMassFunctionToAligneeDatasetFeatures(umcLights);
progData.Report(100);
// Find min/max scan for meta-data
var minScanBaseline = int.MaxValue;
var maxScanBaseline = int.MinValue;
foreach (var feature in umcLights)
{
maxScanBaseline = Math.Max(maxScanBaseline, feature.Scan);
minScanBaseline = Math.Min(minScanBaseline, feature.Scan);
}
// Update the scan and NET ranges
alignmentData.MinScanBaseline = minScanBaseline;
alignmentData.MaxScanBaseline = maxScanBaseline;
alignmentData.MinMTDBNET = (float)alignmentProcessor.MinReferenceNet;
alignmentData.MaxMTDBNET = (float)alignmentProcessor.MaxReferenceNet;
// Cache the matching features
alignmentData.FeatureMatches = alignmentProcessor.FeatureMatches;
// Pull out the heat maps...
OnStatus("Retrieving alignment data.");
progData.Status = "Retrieving alignment data.";
alignmentData.HeatScores = alignmentProcessor.GetAlignmentHeatMap(alignmentOptions.StandardizeHeatScores);
// Mass and net error histograms!
alignmentData.MassErrorHistogram = alignmentProcessor.GetMassErrorHistogram(alignmentOptions.MassBinSize);
alignmentData.NetErrorHistogram = alignmentProcessor.GetNetErrorHistogram(alignmentOptions.NetBinSize);
alignmentData.DriftErrorHistogram = alignmentProcessor.GetDriftErrorHistogram(alignmentOptions.DriftTimeBinSize);
// Get the residual data from the warp.
alignmentData.ResidualData = alignmentProcessor.GetResidualData();
alignmentData.NETIntercept = alignmentProcessor.NetIntercept;
alignmentData.NETRsquared = alignmentProcessor.NetRsquared;
alignmentData.NETSlope = alignmentProcessor.NetSlope;
alignmentData.MassMean = alignmentProcessor.MassMu;
alignmentData.MassStandardDeviation = alignmentProcessor.MassStd;
alignmentData.NETMean = alignmentProcessor.NetMu;
alignmentData.NETStandardDeviation = alignmentProcessor.NetStd;
alignmentData.BaselineIsAmtDB = _options.AlignToMassTagDatabase;
return(alignmentData);
}
/// <summary>
/// Load a single dataset from the provider.
/// </summary>
/// <returns></returns>
public IList <UMCLight> LoadDataset(DatasetInformation dataset,
MsFeatureFilteringOptions msFilteringOptions,
LcmsFeatureFindingOptions lcmsFindingOptions,
LcmsFeatureFilteringOptions lcmsFilteringOptions,
DataLoadingOptions dataLoadOptions,
ScanSummaryProviderCache providerCache,
IdentificationProviderCache identificationProviders,
IProgress <ProgressData> progress = null)
{
var progData = new ProgressData(progress);
IScanSummaryProvider provider = null;
if (!string.IsNullOrWhiteSpace(dataset.RawFile.Path))
{
UpdateStatus("Using raw data to create better features.");
provider = providerCache.GetScanSummaryProvider(dataset.RawFile.Path, dataset.DatasetId);
}
progData.StepRange(1);
progData.Status = "Looking for existing features in the database.";
UpdateStatus(string.Format("[{0}] - Loading dataset [{0}] - {1}.", dataset.DatasetId, dataset.DatasetName));
var datasetId = dataset.DatasetId;
var features = UmcLoaderFactory.LoadUmcFeatureData(dataset, Providers.FeatureCache, provider);
var hasMsFeatures = features.Any(f => f.MsFeatures.Any());
var msFeatures = new List <MSFeatureLight>();
if (!hasMsFeatures)
{
progData.StepRange(2);
progData.Status = "Loading MS Feature Data.";
UpdateStatus(string.Format("[{0}] Loading MS Feature Data [{0}] - {1}.", dataset.DatasetId,
dataset.DatasetName));
var isosFilterOptions = dataLoadOptions.GetIsosFilterOptions();
msFeatures = UmcLoaderFactory.LoadMsFeatureData(dataset.Features.Path, isosFilterOptions);
}
progData.StepRange(3);
progData.Status = "Loading scan summaries.";
////var scansInfo = UmcLoaderFactory.LoadScanSummaries(dataset.Scans.Path);
////dataset.BuildScanTimes(scansInfo);
progData.StepRange(100);
var msnSpectra = new List <MSSpectra>();
// If we don't have any features, then we have to create some from the MS features
// provided to us.
if (features.Count < 1)
{
msFeatures = LcmsFeatureFilters.FilterMsFeatures(msFeatures, msFilteringOptions);
msFeatures = Filter(msFeatures, provider, ref dataset);
progData.Status = "Creating LCMS features.";
features = CreateLcmsFeatures(dataset,
msFeatures,
lcmsFindingOptions,
lcmsFilteringOptions,
provider,
new Progress <ProgressData>(pd => progData.Report(pd.Percent)));
//var maxScan = Convert.ToDouble(features.Max(feature => feature.Scan));
//var minScan = Convert.ToDouble(features.Min(feature => feature.Scan));
var maxScan = features.Max(feature => feature.Scan);
var minScan = features.Min(feature => feature.Scan);
var id = 0;
var scanTimes = dataset.ScanTimes;
foreach (var feature in features)
{
feature.Id = id++;
//feature.Net = (Convert.ToDouble(feature.Scan) - minScan) / (maxScan - minScan);
feature.Net = (Convert.ToDouble(scanTimes[feature.Scan]) - scanTimes[minScan]) / (scanTimes[maxScan] - scanTimes[minScan]);
feature.MassMonoisotopicAligned = feature.MassMonoisotopic;
feature.NetAligned = feature.Net;
feature.GroupId = datasetId;
feature.SpectralCount = feature.MsFeatures.Count;
foreach (var msFeature in feature.MsFeatures.Where(msFeature => msFeature != null))
{
msFeature.UmcId = feature.Id;
msFeature.GroupId = datasetId;
msFeature.MSnSpectra.ForEach(x => x.GroupId = datasetId);
msnSpectra.AddRange(msFeature.MSnSpectra);
}
}
}
else
{
if (!UmcLoaderFactory.AreExistingFeatures(dataset.Features.Path))
{
var i = 0;
foreach (var feature in features)
{
feature.GroupId = datasetId;
feature.Id = i++;
}
}
// Otherwise, we need to map the MS features to the LCMS Features provided.
// This would mean that we extracted data from an existing database.
if (msFeatures.Count > 0)
{
var map = FeatureDataConverters.MapFeature(features);
foreach (var feature in
from feature in msFeatures
let doesFeatureExists = map.ContainsKey(feature.UmcId)
where doesFeatureExists
select feature)
{
map[feature.UmcId].AddChildFeature(feature);
}
}
}
//if (provider is ISpectraProvider)
//{
// var spectraProvider = provider as ISpectraProvider;
// UmcLoaderFactory.LoadMsMs(features.ToList(), spectraProvider);
//}
// Process the MS/MS data with peptides
UpdateStatus("Reading List of Peptides");
if (dataset.SequenceFile != null && !string.IsNullOrEmpty(dataset.SequenceFile.Path))
{
UpdateStatus("Reading List of Peptides");
var idProvider = identificationProviders.GetProvider(dataset.SequenceFile.Path, dataset.DatasetId);
var peptideList = idProvider.GetAllIdentifications();
UpdateStatus("Linking MS/MS to any known Peptide/Metabolite Sequences");
var linker = new PeptideMsMsLinker();
linker.LinkPeptidesToSpectra(msnSpectra, peptideList);
}
progData.Report(100);
return(features);
}
/// <summary>
/// Loads baseline data for alignment.
/// </summary>
private IList <UMCLight> LoadBaselineData(DatasetInformation baselineInfo,
MsFeatureFilteringOptions msFilterOptions,
LcmsFeatureFindingOptions lcmsFindingOptions,
LcmsFeatureFilteringOptions lcmsFilterOptions,
DataLoadingOptions dataLoadOptions,
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,
dataLoadOptions,
m_scanSummaryProviderCache,
this.m_identificationsProvider);
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);
}
