/// <summary>
/// Create XICs for clusters. (Each XIC point becomes a new MSFeature,
// replacing the previous MSFeatures).
/// </summary>
/// <param name="progress">The progress reporter.</param>
/// <param name="umcLights"></param>
/// <returns></returns>
private List <UMCLight> CreateXics(List <UMCLight> umcLights, IProgress <ProgressData> progress)
{
var xicCreator = new XicCreator {
XicRefiner = new XicRefiner(this.options.XicRelativeIntensityThreshold)
};
return(xicCreator.CreateXicNew(
umcLights,
this.options.InstrumentTolerances.Mass,
provider as InformedProteomicsReader,
this.options.RefineXics,
progress).ToList());
}
/// <summary>
/// Finds LCMS Features from MS Features.
/// </summary>
/// <param name="rawMsFeatures"></param>
/// <returns></returns>
public List <TParentFeature> Cluster(List <TChildFeature> rawMsFeatures)
{
Comparison <TChildFeature> mzSort = (x, y) => x.Mz.CompareTo(y.Mz);
Comparison <TParentFeature> monoSort = (x, y) => x.MassMonoisotopic.CompareTo(y.MassMonoisotopic);
Func <TChildFeature, TChildFeature, double> mzDiff = (x, y) => FeatureLight.ComputeMassPPMDifference(x.Mz, y.Mz);
Func <TParentFeature, TParentFeature, double> monoDiff = (x, y) => FeatureLight.ComputeMassPPMDifference(x.MassMonoisotopic, y.MassMonoisotopic);
var minScan = Convert.ToDouble(rawMsFeatures.Min(x => x.Scan));
var maxScan = Convert.ToDouble(rawMsFeatures.Max(x => x.Scan));
foreach (var msFeature in rawMsFeatures)
{
msFeature.Net = (Convert.ToDouble(msFeature.Scan) - minScan) / (maxScan - minScan);
}
OnProgress("Filtering ambiguous features");
//rawMsFeatures = FilterMsFeatures(rawMsFeatures);
OnProgress("Clustering child features into potential UMC candidates");
// First cluster based on m/z finding the XIC's
var features = Cluster <TChildFeature, TParentFeature>(rawMsFeatures,
mzSort,
mzDiff,
CompareMz,
Tolerances.Mass);
var n = features.Count();
OnProgress(string.Format("Found {0} unique child features from {1} total features",
n,
rawMsFeatures.Count()));
OnProgress("Filtering Features");
// Then we group into UMC's for clustering across charge states...
if (features == null)
{
throw new InvalidDataException("No features were found from the input MS Feature list.");
}
OnProgress("Filtering poor features with no data. Calculating statistics for the good ones.");
features = features.Where(x => x.MsFeatures.Count > 0).ToList();
foreach (var feature in features)
{
feature.CalculateStatistics(ClusterCentroidRepresentation.Median);
feature.MassMonoisotopic = (feature.Mz * feature.ChargeState) - (SubAtomicParticleLibrary.MASS_PROTON * feature.ChargeState);
}
// Here we should merge the XIC data...trying to find the best possible feature
// Note that at this point we dont have UMC's. We only have features
// that are separated by mass , scan , and charge
// so this method should interrogate each one of these....
if (SpectraProvider != null)
{
OnProgress(string.Format("Building XIC's from child features"));
var generator = new XicCreator();
generator.Progress += generator_Progress;
features = generator.CreateXic(features as List <UMCLight>, Tolerances.Mass, SpectraProvider) as IEnumerable <TParentFeature>;
generator.Progress -= generator_Progress;
}
OnProgress(string.Format("Calculating statistics for each feature"));
foreach (var feature in features)
{
feature.CalculateStatistics(ClusterCentroidRepresentation.Median);
feature.Net = Convert.ToDouble(feature.Scan - minScan) / Convert.ToDouble(maxScan - minScan);
}
OnProgress(string.Format("Combining child feature charge states"));
features = Cluster <TParentFeature, TParentFeature>(features,
monoSort,
monoDiff,
CompareMonoisotopic,
Tolerances.Mass);
var id = 0;
OnProgress(string.Format("Assigning unique feature id's to each feature."));
var featureList = features.ToList();
foreach (var feature in featureList)
{
feature.Id = id++;
}
return(featureList);
}
