public CrosslinkSearchEngine(List <CrosslinkSpectralMatch>[] globalCsms, Ms2ScanWithSpecificMass[] listOfSortedms2Scans, List <PeptideWithSetModifications> peptideIndex,
List <int>[] fragmentIndex, List <int>[] secondFragmentIndex, int currentPartition, CommonParameters commonParameters, Crosslinker crosslinker, bool CrosslinkSearchTop, int CrosslinkSearchTopNum,
bool quench_H2O, bool quench_NH2, bool quench_Tris, List <string> nestedIds)
: base(null, listOfSortedms2Scans, peptideIndex, fragmentIndex, currentPartition, commonParameters, new OpenSearchMode(), 0, nestedIds)
{
this.GlobalCsms = globalCsms;
this.Crosslinker = crosslinker;
this.CrosslinkSearchTopN = CrosslinkSearchTop;
this.TopN = CrosslinkSearchTopNum;
this.QuenchH2O = quench_H2O;
this.QuenchNH2 = quench_NH2;
this.QuenchTris = quench_Tris;
SecondFragmentIndex = secondFragmentIndex;
if (CommonParameters.ChildScanDissociationType != DissociationType.Unknown && DissociationTypeGenerateSameTypeOfIons(CommonParameters.DissociationType, CommonParameters.ChildScanDissociationType))
{
SecondFragmentIndex = FragmentIndex;
}
GenerateCrosslinkModifications(crosslinker);
AllCrosslinkerSites = Crosslinker.CrosslinkerModSites.ToCharArray().Concat(Crosslinker.CrosslinkerModSites2.ToCharArray()).Distinct().ToArray();
if (commonParameters.PrecursorMassTolerance is PpmTolerance)
{
XLPrecusorSearchMode = new SinglePpmAroundZeroSearchMode(commonParameters.PrecursorMassTolerance.Value);
}
else
{
XLPrecusorSearchMode = new SingleAbsoluteAroundZeroSearchMode(commonParameters.PrecursorMassTolerance.Value);
}
}
/// <summary>
///
/// </summary>
private void GenerateCrosslinkModifications(Crosslinker crosslinker)
{
ModificationMotif.TryGetMotif("X", out var motif);
TrisDeadEnd = new Modification(_originalId: "Tris Dead End", _modificationType: "Crosslink", _locationRestriction: "Anywhere.", _target: motif, _monoisotopicMass: Crosslinker.DeadendMassTris);
H2ODeadEnd = new Modification(_originalId: "H2O Dead End", _modificationType: "Crosslink", _locationRestriction: "Anywhere.", _target: motif, _monoisotopicMass: Crosslinker.DeadendMassH2O);
NH2DeadEnd = new Modification(_originalId: "NH2 Dead End", _modificationType: "Crosslink", _locationRestriction: "Anywhere.", _target: motif, _monoisotopicMass: Crosslinker.DeadendMassNH2);
Loop = new Modification(_originalId: "Loop", _modificationType: "Crosslink", _locationRestriction: "Anywhere.", _target: motif, _monoisotopicMass: Crosslinker.LoopMass);
}
private void PopulateChoices()
{
foreach (var crosslinker in GlobalVariables.Crosslinkers)
{
cbCrosslinkers.Items.Add(crosslinker);
}
Crosslinker DSSO = GlobalVariables.Crosslinkers.First();
cbCrosslinkers.SelectedItem = DSSO;
foreach (string dissassociationType in GlobalVariables.AllSupportedDissociationTypes.Keys)
{
DissociationTypeComboBox.Items.Add(dissassociationType);
ChildScanDissociationTypeComboBox.Items.Add(dissassociationType);
}
cbbXLprecusorMsTl.Items.Add("Da");
cbbXLprecusorMsTl.Items.Add("ppm");
foreach (Protease protease in ProteaseDictionary.Dictionary.Values)
{
proteaseComboBox.Items.Add(protease);
}
Protease trypsin = ProteaseDictionary.Dictionary["trypsin"];
proteaseComboBox.SelectedItem = trypsin;
foreach (string initiatior_methionine_behavior in Enum.GetNames(typeof(InitiatorMethionineBehavior)))
{
initiatorMethionineBehaviorComboBox.Items.Add(initiatior_methionine_behavior);
}
productMassToleranceComboBox.Items.Add("Da");
productMassToleranceComboBox.Items.Add("ppm");
foreach (var hm in GlobalVariables.AllModsKnown.GroupBy(b => b.ModificationType))
{
var theModType = new ModTypeForTreeView(hm.Key, false);
FixedModTypeForTreeViewObservableCollection.Add(theModType);
foreach (var uah in hm)
{
theModType.Children.Add(new ModForTreeView(uah.ToString(), false, uah.IdWithMotif, false, theModType));
}
}
fixedModsTreeView.DataContext = FixedModTypeForTreeViewObservableCollection;
foreach (var hm in GlobalVariables.AllModsKnown.GroupBy(b => b.ModificationType))
{
var theModType = new ModTypeForTreeView(hm.Key, false);
VariableModTypeForTreeViewObservableCollection.Add(theModType);
foreach (var uah in hm)
{
theModType.Children.Add(new ModForTreeView(uah.ToString(), false, uah.IdWithMotif, false, theModType));
}
}
variableModsTreeView.DataContext = VariableModTypeForTreeViewObservableCollection;
}
private void SaveButton_Click(object sender, RoutedEventArgs e)
{
string crosslinkerPath = Path.Combine(GlobalVariables.DataDir, @"Data");
string customCrosslinkPath = Path.Combine(crosslinkerPath, @"CustomCrosslinkers.tsv");
List <string> customCrosslinkerText = new List <string>();
if (File.Exists(customCrosslinkPath))
{
customCrosslinkerText = File.ReadAllLines(customCrosslinkPath).ToList();
}
else
{
customCrosslinkerText.Add("Name\tCrosslinkAminoAcid\tCrosslinkerAminoAcid2\tCleavable\tDissociationType\tCrosslinkerTotalMass\tCrosslinkerShortMass\tCrosslinkerLongMass\tQuenchMassH2O\tQuenchMassNH2\tQuenchMassTris");
}
string name = txtUdXLKerName.Text;
string aminoAcid1 = txtUdXLkerAminoAcids.Text;
string aminoAcid2 = txtUdXLkerAminoAcids2.Text;
bool isCleavable = ckbUdXLkerCleavable.IsChecked.Value;
string dissociationTypes = txtDissociationType.Text;
double mass = double.Parse(txtUdXLkerTotalMs.Text == "" ? "0" : txtUdXLkerTotalMs.Text);
double shortMass = double.Parse(txtUdXLkerShortMass.Text == "" ? "0" : txtUdXLkerShortMass.Text);
double longMass = double.Parse(txtUdXLkerLongMass.Text == "" ? "0" : txtUdXLkerLongMass.Text);
double H2OQuenchMass = double.Parse(txtH2OQuenchMass.Text == "" ? "0" : txtH2OQuenchMass.Text);
double NH2QuenchMass = double.Parse(txtNH2QuenchMass.Text == "" ? "0" : txtNH2QuenchMass.Text);
double TrisQuenchMass = double.Parse(txtTrisQuenchMass.Text == "" ? "0" : txtTrisQuenchMass.Text);
if (GlobalVariables.Crosslinkers.Any(p => p.CrosslinkerName.Contains(name)))
{
MessageBox.Show("A crosslinker already exists with the name: " + name, "Error", MessageBoxButton.OK, MessageBoxImage.Hand);
return;
}
var newCrosslinker = new Crosslinker(crosslinkerName: name, crosslinkerModSites: aminoAcid1, crosslinkerModSites2: aminoAcid2, totalMass: mass, cleavable: isCleavable, dissociationTypes: dissociationTypes,
cleaveMassShort: shortMass, cleaveMassLong: longMass, loopMass: mass, deadendMassH2O: H2OQuenchMass, deadendMassNH2: NH2QuenchMass, deadendMassTris: TrisQuenchMass);
customCrosslinkerText.Add(newCrosslinker.ToString(true));
try
{
File.Delete(customCrosslinkPath);
File.WriteAllLines(customCrosslinkPath, customCrosslinkerText);
}
catch (Exception ex)
{
MessageBox.Show("Problem saving custom crosslinker to file: " + ex.Message, "Error", MessageBoxButton.OK, MessageBoxImage.Hand);
return;
}
GlobalVariables.AddCrosslinkers(new List <Crosslinker> {
newCrosslinker
});
DialogResult = true;
}
public static void XlTestXlPosCal()
{
var prot = new Protein("MNNNKQQQQ", null);
List <DigestionMotif> motifs = new List <DigestionMotif> {
new DigestionMotif("K", null, 1, null)
};
Protease protease = new Protease("New Custom Protease", CleavageSpecificity.Full, null, null, motifs);
ProteaseDictionary.Dictionary.Add(protease.Name, protease);
DigestionParams digestionParams = new DigestionParams(protease: protease.Name, minPeptideLength: 1, initiatorMethionineBehavior: InitiatorMethionineBehavior.Retain);
List <Modification> variableModifications = new List <Modification>();
var ye = prot.Digest(digestionParams, new List <Modification>(), variableModifications).ToList();
var pep = ye[0];
Assert.AreEqual(pep.BaseSequence, "MNNNK");
Crosslinker crosslinker = GlobalVariables.Crosslinkers.Where(p => p.CrosslinkerName == "DSS").First();
Assert.AreEqual(crosslinker.CrosslinkerModSites, "K");
Assert.AreEqual(Residue.GetResidue(crosslinker.CrosslinkerModSites).MonoisotopicMass, 128.09496301518999, 1e-9);
var n = pep.Fragment(DissociationType.HCD, FragmentationTerminus.N);
var c = pep.Fragment(DissociationType.HCD, FragmentationTerminus.C);
Assert.AreEqual(n.Count(), 4);
Assert.AreEqual(c.Count(), 4);
Assert.AreEqual(c.First().NeutralMass, 146.10552769899999, 1e-6);
var x = CrosslinkSpectralMatch.GetPossibleCrosslinkerModSites(crosslinker.CrosslinkerModSites.ToCharArray(), pep).ToArray();
Assert.AreEqual(x[0], 5);
var pep2 = ye[2];
Assert.AreEqual("MNNNKQQQQ", pep2.BaseSequence);
var n2 = pep2.Fragment(DissociationType.HCD, FragmentationTerminus.N);
var c2 = pep2.Fragment(DissociationType.HCD, FragmentationTerminus.C);
Assert.AreEqual(n2.Count(), 8);
Assert.AreEqual(c2.Count(), 8);
var x2 = CrosslinkSpectralMatch.GetPossibleCrosslinkerModSites(crosslinker.CrosslinkerModSites.ToCharArray(), pep2).ToArray();
Assert.AreEqual(x2[0], 5);
//Test crosslinker with multiple types of mod
var protSTC = new Protein("GASTACK", null);
var peps = protSTC.Digest(digestionParams, new List <Modification>(), variableModifications).ToList();
var pepSTC = peps[0];
Assert.AreEqual(pepSTC.BaseSequence, "GASTACK");
Crosslinker crosslinker2 = new Crosslinker("ST", "C", "crosslinkerSTC", false, -18.01056, 0, 0, 0, 0, 0, 0);
string crosslinkerModSitesAll = new string((crosslinker2.CrosslinkerModSites + crosslinker2.CrosslinkerModSites2).ToCharArray().Distinct().ToArray());
Assert.AreEqual(crosslinkerModSitesAll, "STC");
}
/// <summary>
/// Localizes the crosslink position on the alpha and beta peptides
/// </summary>
private CrosslinkSpectralMatch LocalizeCrosslinkSites(Ms2ScanWithSpecificMass theScan, BestPeptideScoreNotch alphaPeptide, BestPeptideScoreNotch betaPeptide, Crosslinker crosslinker, int ind, int inx)
{
CrosslinkSpectralMatch localizedCrosslinkedSpectralMatch = null;
List <Tuple <List <int>, List <int> > > pairs = new List <Tuple <List <int>, List <int> > >();
if (crosslinker.CrosslinkerModSites.Equals(crosslinker.CrosslinkerModSites2))
{
List <int> possibleAlphaXlSites = CrosslinkSpectralMatch.GetPossibleCrosslinkerModSites(crosslinker.CrosslinkerModSites.ToCharArray(), alphaPeptide.BestPeptide);
List <int> possibleBetaXlSites = CrosslinkSpectralMatch.GetPossibleCrosslinkerModSites(crosslinker.CrosslinkerModSites.ToCharArray(), betaPeptide.BestPeptide);
pairs.Add(new Tuple <List <int>, List <int> >(possibleAlphaXlSites, possibleBetaXlSites));
}
else
{
List <int> possibleAlphaXlSites = CrosslinkSpectralMatch.GetPossibleCrosslinkerModSites(crosslinker.CrosslinkerModSites.ToCharArray(), alphaPeptide.BestPeptide);
List <int> possibleBetaXlSites = CrosslinkSpectralMatch.GetPossibleCrosslinkerModSites(crosslinker.CrosslinkerModSites2.ToCharArray(), betaPeptide.BestPeptide);
pairs.Add(new Tuple <List <int>, List <int> >(possibleAlphaXlSites, possibleBetaXlSites));
possibleAlphaXlSites = CrosslinkSpectralMatch.GetPossibleCrosslinkerModSites(crosslinker.CrosslinkerModSites2.ToCharArray(), alphaPeptide.BestPeptide);
possibleBetaXlSites = CrosslinkSpectralMatch.GetPossibleCrosslinkerModSites(crosslinker.CrosslinkerModSites.ToCharArray(), betaPeptide.BestPeptide);
pairs.Add(new Tuple <List <int>, List <int> >(possibleAlphaXlSites, possibleBetaXlSites));
}
foreach (var pair in pairs)
{
List <int> possibleAlphaXlSites = pair.Item1;
List <int> possibleBetaXlSites = pair.Item2;
if (possibleAlphaXlSites.Any() && possibleBetaXlSites.Any())
{
int bestAlphaSite = 0;
int bestBetaSite = 0;
List <MatchedFragmentIon> bestMatchedAlphaIons = new List <MatchedFragmentIon>();
List <MatchedFragmentIon> bestMatchedBetaIons = new List <MatchedFragmentIon>();
Dictionary <int, List <MatchedFragmentIon> > bestMatchedChildAlphaIons = new Dictionary <int, List <MatchedFragmentIon> >();
Dictionary <int, List <MatchedFragmentIon> > bestMatchedChildBetaIons = new Dictionary <int, List <MatchedFragmentIon> >();
double bestAlphaLocalizedScore = 0;
double bestBetaLocalizedScore = 0;
var fragmentsForEachAlphaLocalizedPossibility = CrosslinkedPeptide.XlGetTheoreticalFragments(CommonParameters.DissociationType,
Crosslinker, possibleAlphaXlSites, betaPeptide.BestPeptide.MonoisotopicMass, alphaPeptide.BestPeptide).ToList();
foreach (int possibleSite in possibleAlphaXlSites)
{
foreach (var setOfFragments in fragmentsForEachAlphaLocalizedPossibility.Where(v => v.Item1 == possibleSite))
{
var matchedChildAlphaIons = new Dictionary <int, List <MatchedFragmentIon> >();
var matchedIons = MatchFragmentIons(theScan, setOfFragments.Item2, CommonParameters);
double score = CalculatePeptideScore(theScan.TheScan, matchedIons);
// search child scans (MS2+MS3)
foreach (Ms2ScanWithSpecificMass childScan in theScan.ChildScans)
{
var matchedChildIons = ScoreChildScan(theScan, childScan, possibleSite, alphaPeptide, betaPeptide);
if (matchedChildIons == null)
{
continue;
}
matchedChildAlphaIons.Add(childScan.OneBasedScanNumber, matchedChildIons);
//double childScore = CalculatePeptideScore(childScan.TheScan, matchedChildIons);
//score += childScore;
}
if (score > bestAlphaLocalizedScore)
{
bestAlphaLocalizedScore = score;
bestAlphaSite = possibleSite;
bestMatchedAlphaIons = matchedIons;
bestMatchedChildAlphaIons = matchedChildAlphaIons;
}
}
}
var fragmentsForEachBetaLocalizedPossibility = CrosslinkedPeptide.XlGetTheoreticalFragments(CommonParameters.DissociationType,
Crosslinker, possibleBetaXlSites, alphaPeptide.BestPeptide.MonoisotopicMass, betaPeptide.BestPeptide).ToList();
var alphaMz = new HashSet <double>(bestMatchedAlphaIons.Select(p => p.Mz));
foreach (int possibleSite in possibleBetaXlSites)
{
foreach (var setOfFragments in fragmentsForEachBetaLocalizedPossibility.Where(v => v.Item1 == possibleSite))
{
var matchedIons = MatchFragmentIons(theScan, setOfFragments.Item2, CommonParameters);
var matchedChildBetaIons = new Dictionary <int, List <MatchedFragmentIon> >();
// remove any matched beta ions that also matched to the alpha peptide
matchedIons.RemoveAll(p => alphaMz.Contains(p.Mz));
double score = CalculatePeptideScore(theScan.TheScan, matchedIons);
// search child scans (MS2+MS3)
foreach (Ms2ScanWithSpecificMass childScan in theScan.ChildScans)
{
var matchedChildIons = ScoreChildScan(theScan, childScan, possibleSite, betaPeptide, alphaPeptide);
if (matchedChildIons == null)
{
continue;
}
matchedChildBetaIons.Add(childScan.OneBasedScanNumber, matchedChildIons);
//double childScore = CalculatePeptideScore(childScan.TheScan, matchedChildIons);
//score += childScore;
}
if (score > bestBetaLocalizedScore)
{
bestBetaLocalizedScore = score;
bestBetaSite = possibleSite;
bestMatchedBetaIons = matchedIons;
bestMatchedChildBetaIons = matchedChildBetaIons;
}
}
}
if (bestAlphaLocalizedScore < CommonParameters.ScoreCutoff ||
bestBetaLocalizedScore < CommonParameters.ScoreCutoff)
{
return(null);
}
var localizedAlpha = new CrosslinkSpectralMatch(alphaPeptide.BestPeptide, alphaPeptide.BestNotch, bestAlphaLocalizedScore, 0, theScan, alphaPeptide.BestPeptide.DigestionParams, bestMatchedAlphaIons);
var localizedBeta = new CrosslinkSpectralMatch(betaPeptide.BestPeptide, betaPeptide.BestNotch, bestBetaLocalizedScore, 0, theScan, betaPeptide.BestPeptide.DigestionParams, bestMatchedBetaIons);
localizedAlpha.XlRank = new List <int> {
ind, inx
};
localizedAlpha.XLTotalScore = localizedAlpha.Score + localizedBeta.Score;
localizedAlpha.BetaPeptide = localizedBeta;
localizedAlpha.ChildMatchedFragmentIons = bestMatchedChildAlphaIons;
localizedBeta.ChildMatchedFragmentIons = bestMatchedChildBetaIons;
if (crosslinker.Cleavable)
{
//TODO: re-enable intensity ranks
//psmCrossAlpha.ParentIonMaxIntensityRanks = psmCrossAlpha.MatchedFragmentIons.Where(p => p.NeutralTheoreticalProduct.ProductType == ProductType.M).Select(p => p.IntensityRank).ToList();
//localizedAlpha.ParentIonMaxIntensityRanks = new List<int>();
//localizedAlpha.ParentIonExistNum = psmCrossAlpha.ParentIonMaxIntensityRanks.Count;
}
localizedAlpha.CrossType = PsmCrossType.Cross;
localizedCrosslinkedSpectralMatch = localizedAlpha;
localizedCrosslinkedSpectralMatch.LinkPositions = new List <int> {
bestAlphaSite
};
localizedCrosslinkedSpectralMatch.BetaPeptide.LinkPositions = new List <int> {
bestBetaSite
};
}
}
return(localizedCrosslinkedSpectralMatch);
}
public void WritePepXML_xl(List <CrosslinkSpectralMatch> items, List <Protein> proteinList, string databasePath, List <Modification> variableModifications, List <Modification> fixedModifications, List <string> localizeableModificationTypes, string outputFolder, string fileName, List <string> nestedIds)
{
if (!items.Any())
{
return;
}
XmlSerializer _indexedSerializer = new XmlSerializer(typeof(pepXML.Generated.msms_pipeline_analysis));
var _pepxml = new pepXML.Generated.msms_pipeline_analysis();
_pepxml.date = DateTime.Now;
_pepxml.summary_xml = items[0].FullFilePath + ".pep.XML";
string proteaseC = ""; string proteaseNC = "";
foreach (var x in CommonParameters.DigestionParams.Protease.DigestionMotifs.Select(m => m.InducingCleavage))
{
proteaseC += x;
}
foreach (var x in CommonParameters.DigestionParams.Protease.DigestionMotifs.Select(m => m.PreventingCleavage))
{
proteaseNC += x;
}
Crosslinker crosslinker = XlSearchParameters.Crosslinker;
string fileNameNoExtension = Path.GetFileNameWithoutExtension(items[0].FullFilePath);
string filePathNoExtension = Path.ChangeExtension(items[0].FullFilePath, null);
string modSites = crosslinker.CrosslinkerModSites.ToCharArray().Concat(crosslinker.CrosslinkerModSites2.ToCharArray()).Distinct().ToString();
var para = new List <pepXML.Generated.nameValueType>();
{
para.Add(new pepXML.Generated.nameValueType {
name = "threads", value = CommonParameters.MaxThreadsToUsePerFile.ToString()
});
para.Add(new pepXML.Generated.nameValueType {
name = "database", value = databasePath
});
para.Add(new pepXML.Generated.nameValueType {
name = "MS_data_file", value = items[0].FullFilePath
});
para.Add(new pepXML.Generated.nameValueType {
name = "Cross-link precursor Mass Tolerance", value = CommonParameters.PrecursorMassTolerance.ToString()
});
para.Add(new pepXML.Generated.nameValueType {
name = "Cross-linker type", value = crosslinker.CrosslinkerName
});
para.Add(new pepXML.Generated.nameValueType {
name = "Cross-linker mass", value = crosslinker.TotalMass.ToString()
});
para.Add(new pepXML.Generated.nameValueType {
name = "Cross-linker cleavable", value = crosslinker.Cleavable.ToString()
});
para.Add(new pepXML.Generated.nameValueType {
name = "Cross-linker cleavable long mass", value = crosslinker.CleaveMassLong.ToString()
});
para.Add(new pepXML.Generated.nameValueType {
name = "Cross-linker cleavable short mass", value = crosslinker.CleaveMassShort.ToString()
});
para.Add(new pepXML.Generated.nameValueType {
name = "Cross-linker xl site", value = modSites
});
para.Add(new pepXML.Generated.nameValueType {
name = "Generate decoy proteins", value = XlSearchParameters.DecoyType.ToString()
});
para.Add(new pepXML.Generated.nameValueType {
name = "MaxMissed Cleavages", value = CommonParameters.DigestionParams.MaxMissedCleavages.ToString()
});
para.Add(new pepXML.Generated.nameValueType {
name = "Protease", value = CommonParameters.DigestionParams.Protease.Name
});
para.Add(new pepXML.Generated.nameValueType {
name = "Initiator Methionine", value = CommonParameters.DigestionParams.InitiatorMethionineBehavior.ToString()
});
para.Add(new pepXML.Generated.nameValueType {
name = "Max Modification Isoforms", value = CommonParameters.DigestionParams.MaxModificationIsoforms.ToString()
});
para.Add(new pepXML.Generated.nameValueType {
name = "Min Peptide Len", value = CommonParameters.DigestionParams.MinPeptideLength.ToString()
});
para.Add(new pepXML.Generated.nameValueType {
name = "Max Peptide Len", value = CommonParameters.DigestionParams.MaxPeptideLength.ToString()
});
para.Add(new pepXML.Generated.nameValueType {
name = "Product Mass Tolerance", value = CommonParameters.ProductMassTolerance.ToString()
});
para.Add(new pepXML.Generated.nameValueType {
name = "Ions to search", value = String.Join(", ", DissociationTypeCollection.ProductsFromDissociationType[CommonParameters.DissociationType])
});
foreach (var fixedMod in fixedModifications)
{
para.Add(new pepXML.Generated.nameValueType {
name = "Fixed Modifications: " + fixedMod.IdWithMotif, value = fixedMod.MonoisotopicMass.ToString()
});
}
foreach (var variableMod in variableModifications)
{
para.Add(new pepXML.Generated.nameValueType {
name = "Variable Modifications: " + variableMod.IdWithMotif, value = variableMod.MonoisotopicMass.ToString()
});
}
para.Add(new pepXML.Generated.nameValueType {
name = "Localize All Modifications", value = "true"
});
}
_pepxml.msms_run_summary = new pepXML.Generated.msms_pipeline_analysisMsms_run_summary[1]
{
new pepXML.Generated.msms_pipeline_analysisMsms_run_summary
{
base_name = filePathNoExtension,
raw_data_type = "raw",
raw_data = ".mzML",
sample_enzyme = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySample_enzyme()
{
name = CommonParameters.DigestionParams.Protease.Name,
specificity = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySample_enzymeSpecificity[1]
{
new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySample_enzymeSpecificity
{
cut = proteaseC,
no_cut = proteaseNC,
}
}
},
search_summary = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySearch_summary[1]
{
new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySearch_summary
{
base_name = filePathNoExtension,
search_engine_version = GlobalVariables.MetaMorpheusVersion,
precursor_mass_type = pepXML.Generated.massType.monoisotopic,
fragment_mass_type = pepXML.Generated.massType.monoisotopic,
search_id = 1,
search_database = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySearch_summarySearch_database
{
local_path = databasePath,
type = pepXML.Generated.msms_pipeline_analysisMsms_run_summarySearch_summarySearch_databaseType.AA,
},
enzymatic_search_constraint = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySearch_summaryEnzymatic_search_constraint
{
enzyme = CommonParameters.DigestionParams.Protease.Name,
max_num_internal_cleavages = CommonParameters.DigestionParams.MaxMissedCleavages.ToString(),
//min_number_termini = "2"
},
parameter = para.ToArray()
}
},
}
};
_pepxml.msms_run_summary[0].spectrum_query = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_query[items.Count];
var searchHits = new List <pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hit>();
for (int i = 0; i < items.Count; i++)
{
var mods = new List <pepXML.Generated.modInfoDataTypeMod_aminoacid_mass>();
var alphaPeptide = items[i].BestMatchingPeptides.First().Peptide;
foreach (var modification in alphaPeptide.AllModsOneIsNterminus)
{
var mod = new pepXML.Generated.modInfoDataTypeMod_aminoacid_mass
{
mass = modification.Value.MonoisotopicMass.Value,
// convert from one-based to zero-based (N-term is zero in the pepXML output)
position = (modification.Key - 1).ToString()
};
mods.Add(mod);
}
if (items[i].CrossType == PsmCrossType.Single)
{
var searchHit = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hit
{
hit_rank = 1,
peptide = alphaPeptide.BaseSequence,
peptide_prev_aa = alphaPeptide.PreviousAminoAcid.ToString(),
peptide_next_aa = alphaPeptide.NextAminoAcid.ToString(),
protein = alphaPeptide.Protein.Accession,
num_tot_proteins = 1,
calc_neutral_pep_mass = (float)items[i].ScanPrecursorMass,
massdiff = (items[i].ScanPrecursorMass - items[i].PeptideMonisotopicMass.Value).ToString(),
xlink_typeSpecified = true,
xlink_type = pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hitXlink_type.na,
modification_info = new pepXML.Generated.modInfoDataType {
mod_aminoacid_mass = mods.ToArray()
},
search_score = new pepXML.Generated.nameValueType[]
{
new pepXML.Generated.nameValueType {
name = "xlTotalScore", value = items[i].XLTotalScore.ToString()
},
new pepXML.Generated.nameValueType {
name = "Qvalue", value = items[i].FdrInfo.QValue.ToString()
}
},
};
searchHits.Add(searchHit);
}
else if (items[i].CrossType == PsmCrossType.DeadEnd || items[i].CrossType == PsmCrossType.DeadEndH2O || items[i].CrossType == PsmCrossType.DeadEndNH2 || items[i].CrossType == PsmCrossType.DeadEndTris)
{
double crosslinkerDeadEndMass = 0;
switch (items[i].CrossType)
{
case PsmCrossType.DeadEndNH2:
crosslinkerDeadEndMass = crosslinker.DeadendMassNH2;
break;
case PsmCrossType.DeadEndTris:
crosslinkerDeadEndMass = crosslinker.DeadendMassTris;
break;
default:
crosslinkerDeadEndMass = crosslinker.DeadendMassH2O;
break;
}
var mod = new pepXML.Generated.modInfoDataTypeMod_aminoacid_mass {
mass = crosslinkerDeadEndMass, position = items[i].LinkPositions.First().ToString()
};
mods.Add(mod);
var searchHit = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hit
{
hit_rank = 1,
peptide = alphaPeptide.BaseSequence,
peptide_prev_aa = alphaPeptide.PreviousAminoAcid.ToString(),
peptide_next_aa = alphaPeptide.NextAminoAcid.ToString(),
protein = alphaPeptide.Protein.Accession,
num_tot_proteins = 1,
calc_neutral_pep_mass = (float)items[i].ScanPrecursorMass,
massdiff = (items[i].ScanPrecursorMass - items[i].PeptideMonisotopicMass.Value - crosslinkerDeadEndMass).ToString(),
xlink_typeSpecified = true,
xlink_type = pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hitXlink_type.na,
modification_info = new pepXML.Generated.modInfoDataType {
mod_aminoacid_mass = mods.ToArray()
},
search_score = new pepXML.Generated.nameValueType[]
{
new pepXML.Generated.nameValueType {
name = "xlTotalScore", value = items[i].XLTotalScore.ToString()
},
new pepXML.Generated.nameValueType {
name = "Qvalue", value = items[i].FdrInfo.QValue.ToString()
}
},
};
searchHits.Add(searchHit);
}
else if (items[i].CrossType == PsmCrossType.Inter || items[i].CrossType == PsmCrossType.Intra || items[i].CrossType == PsmCrossType.Cross)
{
var betaPeptide = items[i].BetaPeptide.BestMatchingPeptides.First().Peptide;
var modsBeta = new List <pepXML.Generated.modInfoDataTypeMod_aminoacid_mass>();
foreach (var mod in betaPeptide.AllModsOneIsNterminus)
{
var modBeta = new pepXML.Generated.modInfoDataTypeMod_aminoacid_mass
{
mass = mod.Value.MonoisotopicMass.Value,
// convert from one-based to zero-based (N-term is zero in the pepXML output)
position = (mod.Key - 1).ToString()
};
modsBeta.Add(modBeta);
}
var alpha = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hitXlinkLinked_peptide
{
peptide = alphaPeptide.BaseSequence,
peptide_prev_aa = alphaPeptide.PreviousAminoAcid.ToString(),
peptide_next_aa = alphaPeptide.NextAminoAcid.ToString(),
protein = alphaPeptide.Protein.Accession,
num_tot_proteins = 1,
calc_neutral_pep_mass = (float)items[i].PeptideMonisotopicMass.Value,
complement_mass = (float)(items[i].ScanPrecursorMass - alphaPeptide.MonoisotopicMass),
designation = "alpha",
modification_info = new pepXML.Generated.modInfoDataType {
mod_aminoacid_mass = mods.ToArray()
},
xlink_score = new pepXML.Generated.nameValueType[]
{
new pepXML.Generated.nameValueType {
name = "xlscore", value = items[i].XLTotalScore.ToString()
},
new pepXML.Generated.nameValueType {
name = "link", value = items[i].LinkPositions.First().ToString()
},
}
};
var beta = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hitXlinkLinked_peptide
{
peptide = betaPeptide.BaseSequence,
peptide_prev_aa = betaPeptide.PreviousAminoAcid.ToString(),
peptide_next_aa = betaPeptide.NextAminoAcid.ToString(),
protein = betaPeptide.Protein.Accession,
num_tot_proteins = 1,
calc_neutral_pep_mass = (float)betaPeptide.MonoisotopicMass,
complement_mass = (float)(items[i].ScanPrecursorMass - betaPeptide.MonoisotopicMass),
designation = "beta",
modification_info = new pepXML.Generated.modInfoDataType {
mod_aminoacid_mass = modsBeta.ToArray()
},
xlink_score = new pepXML.Generated.nameValueType[]
{
new pepXML.Generated.nameValueType {
name = "xlscore", value = items[i].BetaPeptide.Score.ToString()
},
new pepXML.Generated.nameValueType {
name = "link", value = items[i].BetaPeptide.LinkPositions.First().ToString()
},
}
};
var cross = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hitXlinkLinked_peptide[2] {
alpha, beta
};
var searchHit = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hit
{
hit_rank = 1,
peptide = "-",
peptide_prev_aa = "-",
peptide_next_aa = "-",
protein = "-",
num_tot_proteins = 1,
calc_neutral_pep_mass = (float)items[i].ScanPrecursorMass,
massdiff = (items[i].ScanPrecursorMass - betaPeptide.MonoisotopicMass - alphaPeptide.MonoisotopicMass - crosslinker.TotalMass).ToString(),
xlink_typeSpecified = true,
xlink_type = pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hitXlink_type.xl,
xlink = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hitXlink
{
identifier = crosslinker.CrosslinkerName,
mass = (float)crosslinker.TotalMass,
linked_peptide = cross
},
search_score = new pepXML.Generated.nameValueType[]
{
new pepXML.Generated.nameValueType {
name = "xlTotalScore", value = items[i].XLTotalScore.ToString()
},
new pepXML.Generated.nameValueType {
name = "Qvalue", value = items[i].FdrInfo.QValue.ToString()
}
}
};
searchHits.Add(searchHit);
}
else if (items[i].CrossType == PsmCrossType.Loop)
{
var thePeptide = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hitXlinkLinked_peptide
{
xlink_score = new pepXML.Generated.nameValueType[]
{
new pepXML.Generated.nameValueType {
name = "link", value = items[i].LinkPositions.First().ToString()
},
new pepXML.Generated.nameValueType {
name = "link", value = items[i].LinkPositions[1].ToString()
}
}
};
var cross = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hitXlinkLinked_peptide[1] {
thePeptide
};
var searchHit = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hit
{
hit_rank = 1,
peptide = alphaPeptide.BaseSequence,
peptide_prev_aa = alphaPeptide.PreviousAminoAcid.ToString(),
peptide_next_aa = alphaPeptide.NextAminoAcid.ToString(),
protein = alphaPeptide.Protein.Accession,
num_tot_proteins = 1,
calc_neutral_pep_mass = (float)items[i].ScanPrecursorMass,
massdiff = (items[i].ScanPrecursorMass - alphaPeptide.MonoisotopicMass - crosslinker.LoopMass).ToString(),
xlink_typeSpecified = true,
xlink_type = pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hitXlink_type.loop,
modification_info = new pepXML.Generated.modInfoDataType {
mod_aminoacid_mass = mods.ToArray()
},
xlink = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hitXlink
{
identifier = crosslinker.CrosslinkerName,
mass = (float)crosslinker.TotalMass,
linked_peptide = cross
},
search_score = new pepXML.Generated.nameValueType[]
{
new pepXML.Generated.nameValueType {
name = "xlTotalScore", value = items[i].XLTotalScore.ToString()
},
new pepXML.Generated.nameValueType {
name = "Qvalue", value = items[i].FdrInfo.QValue.ToString()
}
}
};
searchHits.Add(searchHit);
}
}
for (int i = 0; i < items.Count; i++)
{
_pepxml.msms_run_summary[0].spectrum_query[i] = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_query()
{
spectrum = fileNameNoExtension + "." + items[i].ScanNumber.ToString(),
start_scan = Convert.ToUInt32(items[i].ScanNumber),
end_scan = Convert.ToUInt32(items[i].ScanNumber),
precursor_neutral_mass = (float)items[i].ScanPrecursorMass,
assumed_charge = items[i].ScanPrecursorCharge.ToString(),
index = Convert.ToUInt32(i + 1),
retention_time_sec = (float)(items[i].ScanRetentionTime * 60),
search_result = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_result[1]
{
new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_result
{
search_hit = new pepXML.Generated.msms_pipeline_analysisMsms_run_summarySpectrum_querySearch_resultSearch_hit[1]
{
searchHits[i]
}
}
}
};
}
TextWriter writer = new StreamWriter(Path.Combine(outputFolder, fileName + ".pep.XML"));
_indexedSerializer.Serialize(writer, _pepxml);
writer.Close();
FinishedWritingFile(Path.Combine(outputFolder, fileName + ".pep.XML"), nestedIds);
}
public void WriteCrosslinkToTxtForPercolator(List <CrosslinkSpectralMatch> items, string outputFolder, string fileName, Crosslinker crosslinker, List <string> nestedIds)
{
if (items.Count == 0)
{
return;
}
var writtenFile = Path.Combine(outputFolder, fileName + ".txt");
using (StreamWriter output = new StreamWriter(writtenFile))
{
output.WriteLine("SpecId\tLabel\tScannr\tScore\tdScore\tNormRank\tCharge\tMass\tPPM\tLenShort\tLenLong\tLenSum" +
"\tPeptide\tProtein");
foreach (var item in items)
{
if (item.BaseSequence != null && item.BetaPeptide.BaseSequence != null && item.ProteinAccession != null && item.BetaPeptide.ProteinAccession != null)
{
string x = "T"; int label = 1;
if (item.IsDecoy || item.BetaPeptide.IsDecoy)
{
x = "D"; label = -1;
}
output.WriteLine(
x + "-" + item.ScanNumber.ToString(CultureInfo.InvariantCulture) + "-" + item.ScanRetentionTime.ToString(CultureInfo.InvariantCulture)
+ "\t" + label.ToString(CultureInfo.InvariantCulture)
+ "\t" + item.ScanNumber.ToString(CultureInfo.InvariantCulture)
+ "\t" + item.XLTotalScore.ToString(CultureInfo.InvariantCulture)
+ "\t" + item.DeltaScore.ToString(CultureInfo.InvariantCulture)
+ "\t" + (item.XlRank[0] + item.XlRank[1]).ToString(CultureInfo.InvariantCulture)
+ "\t" + item.ScanPrecursorCharge.ToString(CultureInfo.InvariantCulture)
+ "\t" + item.ScanPrecursorMass.ToString(CultureInfo.InvariantCulture)
+ "\t" + ((item.PeptideMonisotopicMass.HasValue && item.BetaPeptide.PeptideMonisotopicMass.HasValue) ? ((item.ScanPrecursorMass - item.BetaPeptide.PeptideMonisotopicMass.Value - item.PeptideMonisotopicMass.Value - crosslinker.TotalMass) / item.ScanPrecursorMass * 1E6).ToString(CultureInfo.InvariantCulture) : "---")
+ "\t" + item.BetaPeptide.BaseSequence.Length.ToString(CultureInfo.InvariantCulture)
+ "\t" + item.BaseSequence.Length.ToString(CultureInfo.InvariantCulture)
+ "\t" + (item.BetaPeptide.BaseSequence.Length + item.BaseSequence.Length).ToString(CultureInfo.InvariantCulture)
+ "\t" + "-." + item.FullSequence + item.LinkPositions.First().ToString(CultureInfo.InvariantCulture) + "--" + item.BetaPeptide.FullSequence + item.BetaPeptide.LinkPositions.First().ToString(CultureInfo.InvariantCulture) + ".-"
+ "\t" + item.BestMatchingPeptides.First().Peptide.Protein.Accession.ToString(CultureInfo.InvariantCulture)
+ "(" + item.XlProteinPos.ToString(CultureInfo.InvariantCulture) + ")"
+ "\t" + item.BetaPeptide.BestMatchingPeptides.First().Peptide.Protein.Accession.ToString(CultureInfo.InvariantCulture)
+ "(" + item.BetaPeptide.XlProteinPos.ToString(CultureInfo.InvariantCulture) + ")"
);
}
}
}
FinishedWritingFile(writtenFile, nestedIds);
}
public static IEnumerable <Tuple <int, List <Product> > > XlGetTheoreticalFragments(DissociationType dissociationType, Crosslinker crosslinker,
List <int> possibleCrosslinkerPositions, double otherPeptideMass, PeptideWithSetModifications peptide)
{
List <double> massesToLocalize = new List <double>();
if (crosslinker.Cleavable && crosslinker.CleaveDissociationTypes.Contains(dissociationType))
{
massesToLocalize.Add(crosslinker.CleaveMassShort);
massesToLocalize.Add(crosslinker.CleaveMassLong);
}
else
{
massesToLocalize.Add(crosslinker.TotalMass + otherPeptideMass);
}
var fragments = new List <Product>();
HashSet <double> masses = new HashSet <double>();
foreach (int crosslinkerPosition in possibleCrosslinkerPositions)
{
List <Product> theoreticalProducts = new List <Product>(); //need a new one each time to pass as a reference, don't clear
masses.Clear();
foreach (double massToLocalize in massesToLocalize)
{
Dictionary <int, Modification> testMods = new Dictionary <int, Modification> {
{ crosslinkerPosition + 1, new Modification(_monoisotopicMass: massToLocalize) }
};
foreach (var mod in peptide.AllModsOneIsNterminus)
{
testMods.Add(mod.Key, mod.Value);
}
var testPeptide = new PeptideWithSetModifications(peptide.Protein, peptide.DigestionParams, peptide.OneBasedStartResidueInProtein,
peptide.OneBasedEndResidueInProtein, peptide.CleavageSpecificityForFdrCategory, peptide.PeptideDescription, peptide.MissedCleavages, testMods, peptide.NumFixedMods);
testPeptide.Fragment(dissociationType, FragmentationTerminus.Both, fragments);
// add fragmentation ions for this crosslinker position guess
foreach (var fragment in fragments)
{
if (!masses.Contains(fragment.NeutralMass))
{
theoreticalProducts.Add(fragment);
masses.Add(fragment.NeutralMass);
}
}
// add signature ions
if (crosslinker.Cleavable)
{
theoreticalProducts.Add(new Product(ProductType.M, FragmentationTerminus.None, peptide.MonoisotopicMass + massToLocalize,
peptide.Length, peptide.Length, 0));
}
}
yield return(new Tuple <int, List <Product> >(crosslinkerPosition, theoreticalProducts));
}
}
public static void XlTest_BSA_DSSO()
{
//Generate parameters
var commonParameters = new CommonParameters(doPrecursorDeconvolution: false, dissociationType: DissociationType.EThcD,
scoreCutoff: 1, digestionParams: new DigestionParams(minPeptideLength: 5), precursorMassTolerance: new PpmTolerance(10));
var xlSearchParameters = new XlSearchParameters();
//Create databases contain two protein.
var proteinList = new List <Protein> {
new Protein("EKVLTSSAR", "Fake01"), new Protein("LSQKFPK", "Fake02")
};
ModificationMotif.TryGetMotif("M", out ModificationMotif motif1);
Modification mod1 = new Modification(_originalId: "Oxidation of M", _modificationType: "Common Variable", _target: motif1, _locationRestriction: "Anywhere.", _monoisotopicMass: 15.99491461957);
ModificationMotif.TryGetMotif("C", out ModificationMotif motif2);
Modification mod2 = new Modification(_originalId: "Carbamidomethyl of C", _modificationType: "Common Fixed", _target: motif2, _locationRestriction: "Anywhere.", _monoisotopicMass: 57.02146372068994);
var variableModifications = new List <Modification>()
{
mod1
};
var fixedModifications = new List <Modification>()
{
mod2
};
var localizeableModifications = new List <Modification>();
//Run index engine
var indexEngine = new IndexingEngine(proteinList, variableModifications, fixedModifications, null, 1, DecoyType.Reverse, commonParameters, 30000, false, new List <FileInfo>(), new List <string>());
var indexResults = (IndexingResults)indexEngine.Run();
var indexedFragments = indexResults.FragmentIndex.Where(p => p != null).SelectMany(v => v).ToList();
Assert.AreEqual(82, indexedFragments.Count);
Assert.AreEqual(3, indexResults.PeptideIndex.Count);
//Get MS2 scans.
var myMsDataFile = new XLTestDataFile();
var listOfSortedms2Scans = MetaMorpheusTask.GetMs2Scans(myMsDataFile, null, new CommonParameters()).OrderBy(b => b.PrecursorMass).ToArray();
//Generate crosslinker, which is DSSO here.
Crosslinker crosslinker = GlobalVariables.Crosslinkers.Where(p => p.CrosslinkerName == "DSSO").First();
CrosslinkSpectralMatch[] possiblePsms = new CrosslinkSpectralMatch[listOfSortedms2Scans.Length];
new CrosslinkSearchEngine(possiblePsms, listOfSortedms2Scans, indexResults.PeptideIndex, indexResults.FragmentIndex, 0, commonParameters, crosslinker, xlSearchParameters.RestrictToTopNHits, xlSearchParameters.CrosslinkSearchTopNum, xlSearchParameters.XlQuench_H2O, xlSearchParameters.XlQuench_NH2, xlSearchParameters.XlQuench_Tris, new List <string> {
}).Run();
var newPsms = possiblePsms.Where(p => p != null).ToList();
foreach (var item in newPsms)
{
item.SetFdrValues(0, 0, 0, 0, 0, 0, 0, 0, 0, false);
}
//Test newPsms
Assert.AreEqual(3, newPsms.Count);
//Test Output
var task = new XLSearchTask();
task.WritePepXML_xl(newPsms, proteinList, null, variableModifications, fixedModifications, null, TestContext.CurrentContext.TestDirectory, "pep.XML", new List <string> {
});
//Test PsmCross.XlCalculateTotalProductMasses
//var psmCrossAlpha = new CrosslinkSpectralMatch(digestedList[1], 0, 0, 0, listOfSortedms2Scans[0], commonParameters.DigestionParams, new List<MatchedFragmentIon>());
//var psmCrossBeta = new CrosslinkSpectralMatch(digestedList[2], 0, 0, 0, listOfSortedms2Scans[0], commonParameters.DigestionParams, new List<MatchedFragmentIon>());
//var linkPos = CrosslinkSpectralMatch.GetPossibleCrosslinkerModSites(crosslinker.CrosslinkerModSites.ToCharArray(), digestedList[1]);
//var productMassesAlphaList = CrosslinkedPeptide.XlGetTheoreticalFragments(DissociationType.EThcD, false, crosslinker, linkPos, digestedList[2].MonoisotopicMass, digestedList[1]);
//Assert.AreEqual(productMassesAlphaList.First().Value.Count, 50); //TO DO: The number here should be manually verified.
File.Delete(@"singlePsms.tsv");
File.Delete(@"pep.XML.pep.xml");
File.Delete(@"allPsms.tsv");
}
