private LibraryRankedSpectrumInfo(SpectrumPeaksInfo info, IsotopeLabelType labelType,
TransitionGroup group, SrmSettings settings,
string lookupSequence, ExplicitMods lookupMods,
IEnumerable <int> charges, IEnumerable <IonType> types,
IEnumerable <int> rankCharges, IEnumerable <IonType> rankTypes,
bool useFilter, bool matchAll, int minPeaks)
{
LabelType = labelType;
// Avoid ReSharper multiple enumeration warning
var rankChargesArray = rankCharges.ToArray();
var rankTypesArray = rankTypes.ToArray();
if (!useFilter)
{
if (charges == null)
{
charges = GetRanked(rankChargesArray, Transition.ALL_CHARGES);
}
if (types == null)
{
types = GetRanked(rankTypesArray, Transition.ALL_TYPES);
}
matchAll = true;
}
RankParams rp = new RankParams
{
sequence = lookupSequence,
precursorCharge = group.PrecursorCharge,
charges = charges ?? rankCharges,
types = types ?? rankTypes,
matchAll = matchAll,
rankCharges = rankChargesArray,
rankTypes = rankTypesArray,
// Precursor isotopes will not be included in MS/MS, if they will be filtered
// from MS1
excludePrecursorIsotopes = settings.TransitionSettings.FullScan.IsEnabledMs,
tranSettings = settings.TransitionSettings
};
// Get necessary mass calculators and masses
var calcMatchPre = settings.GetPrecursorCalc(labelType, lookupMods);
var calcMatch = settings.GetFragmentCalc(labelType, lookupMods);
var calcPredict = settings.GetFragmentCalc(group.LabelType, lookupMods);
if (!string.IsNullOrEmpty(rp.sequence))
{
rp.precursorMz = SequenceMassCalc.GetMZ(calcMatchPre.GetPrecursorMass(rp.sequence), rp.precursorCharge);
rp.massPreMatch = calcMatch.GetPrecursorFragmentMass(rp.sequence);
rp.massesMatch = calcMatch.GetFragmentIonMasses(rp.sequence);
}
else
{
rp.precursorMz = 0.0;
rp.massPreMatch = 0.0;
rp.massesMatch = new double[0, 0];
}
rp.massPrePredict = rp.massPreMatch;
rp.massesPredict = rp.massesMatch;
if (!ReferenceEquals(calcPredict, calcMatch) && !string.IsNullOrEmpty(rp.sequence))
{
rp.massPrePredict = calcPredict.GetPrecursorFragmentMass(rp.sequence);
rp.massesPredict = calcPredict.GetFragmentIonMasses(rp.sequence);
}
// Get values of interest from the settings.
var tranSettings = settings.TransitionSettings;
var predict = tranSettings.Prediction;
var filter = tranSettings.Filter;
var libraries = tranSettings.Libraries;
var instrument = tranSettings.Instrument;
// Get potential losses to all fragments in this peptide
rp.massType = predict.FragmentMassType;
rp.potentialLosses = TransitionGroup.CalcPotentialLosses(rp.sequence,
settings.PeptideSettings.Modifications, lookupMods,
rp.massType);
// Create arrays because ReadOnlyCollection enumerators are too slow
// In some cases these collections must be enumerated for every ion
// allowed in the library specturm.
rp.startFinder = filter.FragmentRangeFirst;
rp.endFinder = filter.FragmentRangeLast;
// Get library settings
Tolerance = libraries.IonMatchTolerance;
rp.tolerance = Tolerance;
rp.pick = tranSettings.Libraries.Pick;
int ionMatchCount = libraries.IonCount;
// If no library filtering will happen, return all rankings for view in the UI
if (!useFilter || rp.pick == TransitionLibraryPick.none)
{
if (rp.pick == TransitionLibraryPick.none)
{
rp.pick = TransitionLibraryPick.all;
}
ionMatchCount = -1;
}
// Get instrument settings
rp.minMz = instrument.MinMz;
rp.maxMz = instrument.MaxMz;
// Get the library spectrum mass-intensity pairs
IList <SpectrumPeaksInfo.MI> listMI = info.Peaks;
// Because sorting and matching observed ions with predicted
// ions appear as bottlenecks in a profiler, a minimum number
// of peaks may be supplied to allow the use of a 2-phase linear
// filter that can significantly reduce the number of peaks
// needing the O(n*log(n)) sorting and the O(n*m) matching.
int len = listMI.Count;
float intensityCutoff = 0;
if (minPeaks != -1)
{
// Start searching for good cut-off at mean intensity.
double totalIntensity = info.Intensities.Sum();
FindIntensityCutoff(listMI, 0, (float)(totalIntensity / len) * 2, minPeaks, 1, ref intensityCutoff, ref len);
}
// Create filtered peak array storing original index for m/z ordering
// to avoid needing to sort to return to this order.
RankedMI[] arrayRMI = new RankedMI[len];
// Detect when m/z values are out of order, and use the expensive sort
// by m/z to correct this.
double lastMz = double.MinValue;
bool sortMz = false;
for (int i = 0, j = 0, lenOrig = listMI.Count; i < lenOrig; i++)
{
SpectrumPeaksInfo.MI mi = listMI[i];
if (mi.Intensity >= intensityCutoff || intensityCutoff == 0)
{
arrayRMI[j] = new RankedMI(mi, j);
j++;
}
if (ionMatchCount == -1)
{
if (mi.Mz < lastMz)
{
sortMz = true;
}
lastMz = mi.Mz;
}
}
// The one expensive sort is used to determine rank order
// by intensity.
Array.Sort(arrayRMI, OrderIntensityDesc);
RankedMI[] arrayResult = new RankedMI[ionMatchCount != -1 ? ionMatchCount : arrayRMI.Length];
foreach (RankedMI rmi in arrayRMI)
{
rmi.CalculateRank(rp);
// If not filtering for only the highest ionMatchCount ranks
if (ionMatchCount == -1)
{
// Put the ranked record back where it started in the
// m/z ordering to avoid a second sort.
arrayResult[rmi.IndexMz] = rmi;
}
// Otherwise, if this ion was ranked, add it to the result array
else if (rmi.Rank > 0)
{
int countRanks = rmi.Rank;
arrayResult[countRanks - 1] = rmi;
// And stop when the array is full
if (countRanks == ionMatchCount)
{
break;
}
}
}
// If not enough ranked ions were found, fill the rest of the results array
if (ionMatchCount != -1)
{
for (int i = rp.Ranked; i < ionMatchCount; i++)
{
arrayResult[i] = RankedMI.EMPTY;
}
}
// If all ions are to be included, and some were found out of order, then
// the expensive full sort by m/z is necesary.
else if (sortMz)
{
Array.Sort(arrayResult, OrderMz);
}
_spectrum = MakeReadOnly(arrayResult);
}
private LibraryRankedSpectrumInfo(SpectrumPeaksInfo info, IsotopeLabelType labelType,
TransitionGroupDocNode groupDocNode, SrmSettings settings,
Target lookupSequence, ExplicitMods lookupMods,
IEnumerable <Adduct> charges, IEnumerable <IonType> types,
IEnumerable <Adduct> rankCharges, IEnumerable <IonType> rankTypes,
double?score, bool useFilter, bool matchAll, int minPeaks)
{
LabelType = labelType;
// Avoid ReSharper multiple enumeration warning
var rankChargesArray = rankCharges.ToArray();
var rankTypesArray = rankTypes.ToArray();
TransitionGroup group = groupDocNode.TransitionGroup;
bool isProteomic = group.IsProteomic;
if (score == null && groupDocNode.HasLibInfo && groupDocNode.LibInfo is BiblioSpecSpectrumHeaderInfo libInfo)
{
Score = libInfo.Score;
}
else
{
Score = score;
}
if (!useFilter)
{
if (charges == null)
{
charges = GetRanked(rankChargesArray, isProteomic ? Transition.DEFAULT_PEPTIDE_CHARGES : Transition.DEFAULT_MOLECULE_CHARGES);
}
if (types == null)
{
types = GetRanked(rankTypesArray, isProteomic ? Transition.PEPTIDE_ION_TYPES : Transition.MOLECULE_ION_TYPES);
}
matchAll = true;
}
bool limitRanks =
groupDocNode.IsCustomIon && // For small molecules, cap the number of ranked ions displayed if we don't have any peak metadata
groupDocNode.Transitions.Any(t => string.IsNullOrEmpty(t.FragmentIonName));
RankParams rp = new RankParams
{
sequence = lookupSequence,
precursorAdduct = group.PrecursorAdduct,
adducts = charges ?? rankCharges,
types = types ?? rankTypes,
matchAll = matchAll,
rankCharges = rankChargesArray.Select(a => Math.Abs(a.AdductCharge)).ToArray(),
rankTypes = rankTypesArray,
// Precursor isotopes will not be included in MS/MS, if they will be filtered
// from MS1
excludePrecursorIsotopes = settings.TransitionSettings.FullScan.IsEnabledMs,
tranSettings = settings.TransitionSettings,
rankLimit = limitRanks ? settings.TransitionSettings.Libraries.IonCount : (int?)null
};
// Get necessary mass calculators and masses
var calcMatchPre = settings.GetPrecursorCalc(labelType, lookupMods);
var calcMatch = isProteomic ? settings.GetFragmentCalc(labelType, lookupMods) : settings.GetDefaultFragmentCalc();
var calcPredict = isProteomic ? settings.GetFragmentCalc(group.LabelType, lookupMods) : calcMatch;
if (isProteomic && rp.sequence.IsProteomic)
{
rp.precursorMz = SequenceMassCalc.GetMZ(calcMatchPre.GetPrecursorMass(rp.sequence), rp.precursorAdduct);
rp.massPreMatch = calcMatch.GetPrecursorFragmentMass(rp.sequence);
rp.massesMatch = calcMatch.GetFragmentIonMasses(rp.sequence);
rp.knownFragments = null;
}
else if (!isProteomic && !rp.sequence.IsProteomic)
{
string isotopicForumla;
rp.precursorMz = SequenceMassCalc.GetMZ(calcMatchPre.GetPrecursorMass(rp.sequence.Molecule, null, rp.precursorAdduct, out isotopicForumla), rp.precursorAdduct);
rp.massPreMatch = calcMatch.GetPrecursorFragmentMass(rp.sequence);
// rp.massesMatch = calcMatch.GetFragmentIonMasses(rp.molecule); CONSIDER, for some molecule types someday?
// For small molecules we can't predict fragmentation, so just use those we have
// Older Resharper code inspection implementations insist on warning here
// Resharper disable PossibleMultipleEnumeration
var existing = groupDocNode.Transitions.Where(tran => tran.Transition.IsNonPrecursorNonReporterCustomIon()).Select(t => t.Transition.CustomIon.GetMass(MassType.Monoisotopic)).ToArray();
rp.massesMatch = new IonTable <TypedMass>(IonType.custom, existing.Length);
for (var i = 0; i < existing.Length; i++)
{
rp.massesMatch[IonType.custom, i] = existing[i];
}
// Resharper restore PossibleMultipleEnumeration
rp.knownFragments = groupDocNode.Transitions.Where(tran => tran.Transition.IsNonPrecursorNonReporterCustomIon()).Select(t =>
new KnownFragment
{
Adduct = t.Transition.Adduct,
Name = t.GetFragmentIonName(CultureInfo.CurrentCulture, settings.TransitionSettings.Libraries.IonMatchTolerance),
Mz = t.Mz
}).ToList();
}
else
{
rp.precursorMz = 0.0;
rp.massPreMatch = TypedMass.ZERO_MONO_MASSH;
rp.massesMatch = IonTable <TypedMass> .EMPTY;
rp.knownFragments = null;
}
rp.massPrePredict = rp.massPreMatch;
rp.massesPredict = rp.massesMatch;
if (!ReferenceEquals(calcPredict, calcMatch))
{
rp.massPrePredict = calcPredict.GetPrecursorFragmentMass(rp.sequence);
if (rp.sequence.IsProteomic) // CONSIDER - eventually we may be able to predict fragments for small molecules?
{
rp.massesPredict = calcPredict.GetFragmentIonMasses(rp.sequence);
}
}
// Get values of interest from the settings.
var tranSettings = settings.TransitionSettings;
var predict = tranSettings.Prediction;
var filter = tranSettings.Filter;
var libraries = tranSettings.Libraries;
var instrument = tranSettings.Instrument;
// Get potential losses to all fragments in this peptide
rp.massType = predict.FragmentMassType;
rp.potentialLosses = TransitionGroup.CalcPotentialLosses(rp.sequence,
settings.PeptideSettings.Modifications, lookupMods,
rp.massType);
// Create arrays because ReadOnlyCollection enumerators are too slow
// In some cases these collections must be enumerated for every ion
// allowed in the library specturm.
rp.startFinder = filter.FragmentRangeFirst;
rp.endFinder = filter.FragmentRangeLast;
// Get library settings
Tolerance = libraries.IonMatchTolerance;
rp.tolerance = Tolerance;
rp.pick = tranSettings.Libraries.Pick;
int ionMatchCount = libraries.IonCount;
// If no library filtering will happen, return all rankings for view in the UI
if (!useFilter || rp.pick == TransitionLibraryPick.none)
{
if (rp.pick == TransitionLibraryPick.none)
{
rp.pick = TransitionLibraryPick.all;
}
ionMatchCount = -1;
}
// Get instrument settings
rp.minMz = instrument.MinMz;
rp.maxMz = instrument.MaxMz;
// Get the library spectrum mass-intensity pairs
IList <SpectrumPeaksInfo.MI> listMI = info.Peaks;
// Because sorting and matching observed ions with predicted
// ions appear as bottlenecks in a profiler, a minimum number
// of peaks may be supplied to allow the use of a 2-phase linear
// filter that can significantly reduce the number of peaks
// needing the O(n*log(n)) sorting and the O(n*m) matching.
int len = listMI.Count;
float intensityCutoff = 0;
if (minPeaks != -1)
{
// Start searching for good cut-off at mean intensity.
double totalIntensity = info.Intensities.Sum();
FindIntensityCutoff(listMI, 0, (float)(totalIntensity / len) * 2, minPeaks, 1, ref intensityCutoff, ref len);
}
// Create filtered peak array storing original index for m/z ordering
// to avoid needing to sort to return to this order.
RankedMI[] arrayRMI = new RankedMI[len];
// Detect when m/z values are out of order, and use the expensive sort
// by m/z to correct this.
double lastMz = double.MinValue;
bool sortMz = false;
for (int i = 0, j = 0, lenOrig = listMI.Count; i < lenOrig; i++)
{
SpectrumPeaksInfo.MI mi = listMI[i];
if (mi.Intensity >= intensityCutoff || intensityCutoff == 0)
{
arrayRMI[j] = new RankedMI(mi, j);
j++;
}
if (ionMatchCount == -1)
{
if (mi.Mz < lastMz)
{
sortMz = true;
}
lastMz = mi.Mz;
}
}
// The one expensive sort is used to determine rank order
// by intensity, or m/z in case of a tie.
Array.Sort(arrayRMI, OrderIntensityDesc);
RankedMI[] arrayResult = new RankedMI[ionMatchCount != -1 ? ionMatchCount : arrayRMI.Length];
foreach (RankedMI rmi in arrayRMI)
{
rmi.CalculateRank(rp);
// If not filtering for only the highest ionMatchCount ranks
if (ionMatchCount == -1)
{
// Put the ranked record back where it started in the
// m/z ordering to avoid a second sort.
arrayResult[rmi.IndexMz] = rmi;
}
// Otherwise, if this ion was ranked, add it to the result array
else if (rmi.Rank > 0)
{
int countRanks = rmi.Rank;
arrayResult[countRanks - 1] = rmi;
// And stop when the array is full
if (countRanks == ionMatchCount)
{
break;
}
}
}
// Is this a theoretical library with no intensity variation? If so it can't be ranked.
// If it has any interesting peak annotations, pass those through
if (rp.Ranked == 0 && arrayRMI.All(rmi => rmi.Intensity == arrayRMI[0].Intensity))
{
// Only do this if we have been asked to limit the ions matched, and there are any annotations
if (ionMatchCount != -1 && arrayRMI.Any(rmi => rmi.HasAnnotations))
{
// Pass through anything with an annotation as being of probable interest
arrayResult = arrayRMI.Where(rmi => rmi.HasAnnotations).ToArray();
ionMatchCount = -1;
}
}
// If not enough ranked ions were found, fill the rest of the results array
if (ionMatchCount != -1)
{
for (int i = rp.Ranked; i < ionMatchCount; i++)
{
arrayResult[i] = RankedMI.EMPTY;
}
}
// If all ions are to be included, and some were found out of order, then
// the expensive full sort by m/z is necesary.
else if (sortMz)
{
Array.Sort(arrayResult, OrderMz);
}
_spectrum = MakeReadOnly(arrayResult);
}
public SpectrumRanker(TargetInfo targetInfo, SrmSettings settings,
FragmentFilter fragmentFilter)
{
TargetInfoObj = targetInfo;
FragmentFilterObj = fragmentFilter;
var groupDocNode = TargetInfoObj.TransitionGroupDocNode;
TransitionGroup group = groupDocNode.TransitionGroup;
bool isProteomic = group.IsProteomic;
bool limitRanks =
groupDocNode.IsCustomIon && // For small molecules, cap the number of ranked ions displayed if we don't have any peak metadata
groupDocNode.Transitions.Any(t => string.IsNullOrEmpty(t.FragmentIonName));
RankLimit = limitRanks ? settings.TransitionSettings.Libraries.IonCount : (int?)null;
// Get necessary mass calculators and masses
var labelType = targetInfo.SpectrumLabelType;
var lookupMods = targetInfo.LookupMods;
var calcMatchPre = settings.GetPrecursorCalc(labelType, lookupMods);
var calcMatch = isProteomic ? settings.GetFragmentCalc(labelType, lookupMods) : settings.GetDefaultFragmentCalc();
var calcPredict = isProteomic ? settings.GetFragmentCalc(group.LabelType, lookupMods) : calcMatch;
MoleculeMasses moleculeMasses;
if (null != lookupMods && lookupMods.HasCrosslinks)
{
moleculeMasses = GetCrosslinkMasses(settings);
}
else
{
if (isProteomic && Sequence.IsProteomic)
{
moleculeMasses = new MoleculeMasses(
SequenceMassCalc.GetMZ(calcMatchPre.GetPrecursorMass(Sequence), PrecursorAdduct),
new IonMasses(calcMatch.GetPrecursorFragmentMass(Sequence),
calcMatch.GetFragmentIonMasses(Sequence)));
}
else if (!isProteomic && !Sequence.IsProteomic)
{
string isotopicFormula;
var knownFragments = new List <MatchedFragmentIon>();
foreach (var tran in groupDocNode.Transitions)
{
if (tran.Transition.IsNonPrecursorNonReporterCustomIon())
{
knownFragments.Add(new MatchedFragmentIon(IonType.custom, knownFragments.Count + 1,
tran.Transition.Adduct,
tran.GetFragmentIonName(CultureInfo.CurrentCulture,
settings.TransitionSettings.Libraries.IonMatchTolerance),
null,
tran.Mz));
}
}
var ionMasses =
new IonMasses(calcMatch.GetPrecursorFragmentMass(Sequence), IonTable <TypedMass> .EMPTY)
.ChangeKnownFragments(knownFragments);
moleculeMasses =
new MoleculeMasses(
SequenceMassCalc.GetMZ(
calcMatchPre.GetPrecursorMass(Sequence.Molecule, null, PrecursorAdduct,
out isotopicFormula), PrecursorAdduct), ionMasses);
}
else
{
moleculeMasses = new MoleculeMasses(0.0,
new IonMasses(TypedMass.ZERO_MONO_MASSH, IonTable <TypedMass> .EMPTY));
}
if (!ReferenceEquals(calcPredict, calcMatch))
{
var ionTable = moleculeMasses.MatchIonMasses.FragmentMasses;
if (Sequence.IsProteomic
) // CONSIDER - eventually we may be able to predict fragments for small molecules?
{
ionTable = calcPredict.GetFragmentIonMasses(Sequence);
}
moleculeMasses =
moleculeMasses.ChangePredictIonMasses(new IonMasses(
calcPredict.GetPrecursorFragmentMass(Sequence),
ionTable));
}
}
MoleculeMassesObj = moleculeMasses;
// Get values of interest from the settings.
TransitionSettings = settings.TransitionSettings;
// Get potential losses to all fragments in this peptide
PotentialLosses = TransitionGroup.CalcPotentialLosses(Sequence, settings.PeptideSettings.Modifications,
lookupMods, MassType);
}