public static void MatchAllFragments(Result rez)
{
Console.WriteLine(" ************************************************************************ ");
Console.WriteLine(" ****************** Testing All Fragments on results ******************** ");
rez.dbOptions.fragments = new Fragments();
rez.dbOptions.fragments.Add(new FragmentA());
rez.dbOptions.fragments.Add(new FragmentB());
rez.dbOptions.fragments.Add(new FragmentC());
rez.dbOptions.fragments.Add(new FragmentX());
rez.dbOptions.fragments.Add(new FragmentY());
rez.dbOptions.fragments.Add(new FragmentZ());
rez.dbOptions.addFragmentMods = true;
rez.dbOptions.addFragmentLoss = true;
foreach (Precursor precursor in rez.matchedPrecursors)
{
PeptideSpectrumMatch psm = precursor.OptimizedBestPsm();
PeptideSpectrumMatch psmNew = new PeptideSpectrumMatch(psm.Query, psm.Peptide, rez.dbOptions);
precursor.psms = new PeptideSpectrumMatches();// GraphML_List<PeptideSpectrumMatch>();
precursor.psms.Add(psmNew);
}
rez.WriteInfoToCsv();
rez.Export(0.05, "05_AllFragments", true);
Console.WriteLine(" ****************** All fragments added to best psms ******************** ");
Console.WriteLine(" ************************************************************************ ");
}
public static AnnotatedSpectrum TestNeo4j(IConSol console)
{
DBOptions dbOptions = GetDBOptions(false, false, console);
Samples ProjectRatios = new Samples(@"C:\_IRIC\DATA\NB\ProjectTest_MonoAce_Spiked_19Oct.csv", 0, dbOptions);//Group 2 (all)
Result rez = Propheus.Start(dbOptions, ProjectRatios, false, false, false);
Database.Neo4j.ResultsToNeo4j.Export(rez);
PeptideSpectrumMatch bestPsm = null;
foreach (Query query in rez.queries)
{
foreach (PeptideSpectrumMatch psm in query.psms)
{
if (psm.Peptide.Sequence.CompareTo("GK(acetylation of K)GGK(propionylation of K)GLGK(propionylation of K)GGAK(propionylation of K)R") == 0 &&
(bestPsm == null || bestPsm.ProbabilityScore() < query.psms[0].ProbabilityScore()))
{
bestPsm = psm;
}
}
}
AnnotatedSpectrum aSpec = new AnnotatedSpectrum(ProjectRatios[0], bestPsm.Query.spectrum, bestPsm.Peptide);
return(aSpec);
}
void setData(object sender, DoWorkEventArgs e)
{
try
{
Map <int, List <PSMRow> > precursorMassErrorsByCharge = new Map <int, List <PSMRow> >();;
IDictionary <int, int> spectralCountByChargeState;
lock (session)
{
var randomIds = session.CreateQuery("SELECT psm.Id " + viewFilter.GetFilteredQueryString(DataFilter.FromPeptideSpectrumMatch))
.List <long>()
.Shuffle()
.Take(50000)
.OrderBy(o => o);
string randomIdSet = String.Join(",", randomIds.Select(o => o.ToString()).ToArray());
var query = session.CreateSQLQuery("SELECT psm.ObservedNeutralMass, psm.MonoisotopicMassError, psm.MolecularWeightError, s.ScanTimeInSeconds, psm.Charge " +
"FROM PeptideSpectrumMatch psm " +
"JOIN Spectrum s ON psm.Spectrum=s.Id " +
"WHERE psm.Id IN (" + randomIdSet + ") " +
"GROUP BY psm.Id");
query.List <object[]>().ForEach(o =>
precursorMassErrorsByCharge[Convert.ToInt32(o[4])].Add(new PSMRow
{
ObservedNeutralMass = Convert.ToDouble(o[0]),
MassError = PeptideSpectrumMatch.GetSmallerMassError(Convert.ToDouble(o[1]), Convert.ToDouble(o[2])),
ScanTime = Convert.ToDouble(o[3]) / 60, // convert to minutes
Charge = Convert.ToInt32(o[4])
}));
var query2 = session.CreateQuery("SELECT psm.Charge, COUNT(DISTINCT psm.Spectrum.id) " +
dataFilter.GetFilteredQueryString(DataFilter.FromPeptideSpectrumMatch) +
"GROUP BY psm.Charge");
spectralCountByChargeState = query2.List <object[]>().ToDictionary(o => Convert.ToInt32(o[0]), o => Convert.ToInt32(o[1]));
}
Map <int, List <List <PSMRow> > > clusteredPrecursorMassErrorsByCharge = clusterMassErrors(precursorMassErrorsByCharge);
// convert to PPM if the user requested it
if (!ppmMassErrorMenuItem.Text.Contains("PPM"))
{
clusteredPrecursorMassErrorsByCharge.Values.ForEach(o3 => o3.ForEach(o2 => o2.ForEach(o => o.MassError = o.MassError / o.ObservedNeutralMass * 1e6)));
}
{
var csr = new ColorSymbolRotator();
precursorMassErrorForm.ZedGraphControl.GraphPane.CurveList.Clear();
foreach (var kvp in clusteredPrecursorMassErrorsByCharge)
{
bool firstCluster = true;
var color = csr.NextColor;
foreach (var precursorMassErrorCluster in kvp.Value)
{
var precursorMassErrors = precursorMassErrorCluster;
var precursorMassErrorValues = precursorMassErrors.Select(o => o.MassError).ToArray();
var densityCurve = new LineItem(firstCluster ? kvp.Key.ToString() : "", getDensityCurve(precursorMassErrorValues, 0), color, SymbolType.None);
var grassPlot = new LineItem("", precursorMassErrorValues, Enumerable.Repeat(0.0, precursorMassErrorValues.Length).ToArray(), color, SymbolType.VDash)
{
Tag = kvp.Key
};
densityCurve.Line.IsAntiAlias = true;
precursorMassErrorForm.ZedGraphControl.GraphPane.CurveList.Add(densityCurve);
precursorMassErrorForm.ZedGraphControl.GraphPane.CurveList.Add(grassPlot);
firstCluster = false;
}
}
}
{
var csr = new ColorSymbolRotator();
scanTimeDistributionForm.ZedGraphControl.GraphPane.CurveList.Clear();
foreach (var kvp in precursorMassErrorsByCharge)
{
var precursorMassErrors = kvp.Value;
var scanTimeValues = precursorMassErrors.Select(o => o.ScanTime).ToList();
var color = csr.NextColor;
if (scanTimeValues.Max() > 0)
{
var densityCurve = new LineItem(kvp.Key.ToString(), getDensityCurve(scanTimeValues, 0), color, SymbolType.None);
var grassPlot = new LineItem("", scanTimeValues.ToArray(), Enumerable.Repeat(0.0, scanTimeValues.Count).ToArray(), color, SymbolType.VDash);
densityCurve.Line.IsAntiAlias = true;
scanTimeDistributionForm.ZedGraphControl.GraphPane.CurveList.Add(densityCurve);
scanTimeDistributionForm.ZedGraphControl.GraphPane.CurveList.Add(grassPlot);
}
}
if (scanTimeDistributionForm.ZedGraphControl.GraphPane.CurveList.IsNullOrEmpty())
{
scanTimeDistributionForm.ZedGraphControl.GraphPane.GraphObjList.Add(
new TextObj("No scan time information present.\r\nUse File/Embed to import scan times from raw data.",
0.5, 0.5, CoordType.ChartFraction, AlignH.Center, AlignV.Center)
{
FontSpec = new FontSpec("monospace", 18.0f, Color.Black, true, false, false)
});
}
}
var colorRotator = new ColorSymbolRotator();
var colors = new List <Color>();
foreach (var kvp in spectralCountByChargeState)
{
colors.Add(colorRotator.NextColor);
}
chargeStatesForm.ZedGraphControl.GraphPane.CurveList.Clear();
chargeStatesForm.ZedGraphControl.GraphPane.BarSettings.MinClusterGap = 8;
var barItem = chargeStatesForm.ZedGraphControl.GraphPane.AddBar("", spectralCountByChargeState.Keys.Select(o => (double)o).ToArray(), spectralCountByChargeState.Values.Select(o => (double)o).ToArray(), Color.White) as BarItem;
for (int i = 0; i < barItem.Points.Count; ++i)
{
barItem.Points[i].ColorValue = (double)i + 1;
}
barItem.Bar.Fill = new Fill(colors.ToArray())
{
Type = FillType.GradientByColorValue,
RangeMin = 1,
RangeMax = colors.Count
};
chargeStatesForm.ZedGraphControl.GraphPane.XAxis.Scale.TextLabels = spectralCountByChargeState.Keys.Select(o => o.ToString()).ToArray();
}
catch (Exception ex)
{
e.Result = ex;
}
}
public static void CreateTestData (NHibernate.ISession session, IList<SpectrumTuple> testPsmSummary)
{
var dbGroups = new Map<string, SpectrumSourceGroup>();
foreach (var ssg in session.Query<SpectrumSourceGroup>())
dbGroups[ssg.Name] = ssg;
var dbSources = new Map<long, SpectrumSource>();
foreach (var ss in session.Query<SpectrumSource>())
dbSources[ss.Id.Value] = ss;
var dbAnalyses = new Map<long, Analysis>();
foreach (var a in session.Query<Analysis>())
dbAnalyses[a.Id.Value] = a;
var dbPeptides = new Map<string, Peptide>();
foreach (var pep in session.Query<Peptide>())
dbPeptides[pep.Sequence] = pep;
var bulkInserter = new BulkInserter(session.Connection);
long lastPsmId = session.CreateQuery("SELECT MAX(Id) FROM PeptideSpectrumMatch").UniqueResult<long?>().GetValueOrDefault();
long lastModId = session.CreateQuery("SELECT MAX(Id) FROM Modification").UniqueResult<long?>().GetValueOrDefault();
long lastPmId = session.CreateQuery("SELECT MAX(Id) FROM PeptideModification").UniqueResult<long?>().GetValueOrDefault();
long lastGroupId = session.CreateQuery("SELECT MAX(Id) FROM SpectrumSourceGroup").UniqueResult<long?>().GetValueOrDefault();
long lastSourceId = session.CreateQuery("SELECT MAX(Id) FROM SpectrumSource").UniqueResult<long?>().GetValueOrDefault();
long lastSglId = session.CreateQuery("SELECT MAX(Id) FROM SpectrumSourceGroupLink").UniqueResult<long?>().GetValueOrDefault();
foreach (SpectrumTuple row in testPsmSummary)
{
string groupName = row.Group;
string sourceName = "Source " + row.Source;
string analysisId = "Engine " + row.Analysis;
string peptideTuples = row.PeptideTuples;
SpectrumSourceGroup group = dbGroups[groupName];
if (String.IsNullOrEmpty(group.Name))
{
group.Id = ++lastGroupId;
group.Name = groupName;
bulkInserter.Add(group);
}
SpectrumSource source = dbSources[row.Source];
if (String.IsNullOrEmpty(source.Name))
{
source.Id = ++lastSourceId;
source.Name = sourceName;
source.Group = group;
source.Spectra = new List<Spectrum>();
bulkInserter.Add(source);
// add a source group link for the source's immediate group
bulkInserter.Add(new SpectrumSourceGroupLink() { Id = ++lastSglId, Group = group, Source = source });
#region add source group links for all of the immediate group's parent groups
if (groupName != "/")
{
string parentGroupName = groupName.Substring(0, groupName.LastIndexOf("/"));
while (true)
{
if (String.IsNullOrEmpty(parentGroupName))
parentGroupName = "/";
// add the parent group if it doesn't exist yet
SpectrumSourceGroup parentGroup = session.UniqueResult<SpectrumSourceGroup>(o => o.Name == parentGroupName);
if (parentGroup == null)
{
parentGroup = new SpectrumSourceGroup() { Id = ++lastGroupId, Name = parentGroupName };
bulkInserter.Add(parentGroup);
}
bulkInserter.Add(new SpectrumSourceGroupLink() { Id = ++lastSglId, Group = parentGroup, Source = source });
if (parentGroupName == "/")
break;
parentGroupName = parentGroupName.Substring(0, parentGroupName.LastIndexOf("/"));
}
}
#endregion
}
Spectrum spectrum = source.Spectra.SingleOrDefault(o => o.Source.Id == source.Id &&
o.Index == row.Spectrum - 1);
if (spectrum == null)
{
spectrum = new Spectrum()
{
Id = source.Id * 10000 + row.Spectrum,
Index = row.Spectrum - 1,
NativeID = "scan=" + row.Spectrum,
Source = source,
PrecursorMZ = 42
};
source.Spectra.Add(spectrum);
bulkInserter.Add(spectrum);
}
Analysis analysis = dbAnalyses[row.Analysis];
if (String.IsNullOrEmpty(analysis.Name))
{
analysis.Id = dbAnalyses.Max(o => o.Value.Id).GetValueOrDefault() + 1;
analysis.Name = analysisId + " 1.0";
analysis.Software = new AnalysisSoftware() {Name = analysisId, Version = "1.0"};
analysis.StartTime = DateTime.Today.AddHours(row.Analysis);
analysis.Type = AnalysisType.DatabaseSearch;
analysis.Parameters = new SortedSet<AnalysisParameter>()
{
new AnalysisParameter()
{
Id = analysis.Id * 10000,
Analysis = analysis,
Name = "Parameter 1",
Value = "Value 1"
}
};
bulkInserter.Add(analysis);
}
// make sure peptides are sorted by their score divider (which will determine rank)
var peptideList = new SortedList<int, List<PeptideTuple>>();
foreach (string tuple in peptideTuples.Split(" ".ToCharArray(), StringSplitOptions.RemoveEmptyEntries))
{
var peptideTuple = new PeptideTuple()
{
Sequence = tuple.Split('@', '/')[0],
Charge = Convert.ToInt32(tuple.Split('@', '/')[1]),
ScoreDivider = Convert.ToInt32(tuple.Split('@', '/')[2])
};
if (!peptideList.ContainsKey(peptideTuple.ScoreDivider))
peptideList[peptideTuple.ScoreDivider] = new List<PeptideTuple>();
peptideList[peptideTuple.ScoreDivider].Add(peptideTuple);
}
int rank = 1;
int lastDivider = 1;
foreach (var peptideTupleList in peptideList.Values)
foreach (var peptideTuple in peptideTupleList)
{
using (PwizPeptide pwizPeptide = new PwizPeptide(peptideTuple.Sequence, ModParsing.ModificationParsing_Auto, ModDelimiter.ModificationDelimiter_Brackets))
{
Peptide peptide = dbPeptides[pwizPeptide.sequence];
if (String.IsNullOrEmpty(peptide.Sequence))
{
peptide = new TestPeptide(pwizPeptide.sequence);
peptide.Id = dbPeptides.Max(o => o.Value.Id).GetValueOrDefault() + 1;
peptide.MonoisotopicMass = pwizPeptide.monoisotopicMass(false);
peptide.MolecularWeight = pwizPeptide.molecularWeight(false);
dbPeptides[pwizPeptide.sequence] = peptide;
bulkInserter.Add(peptide);
createTestPeptideInstances(session, bulkInserter, peptide);
}
double neutralPrecursorMass = (spectrum.PrecursorMZ*peptideTuple.Charge) - (peptideTuple.Charge*Proton.Mass);
var psm = new PeptideSpectrumMatch()
{
Id = ++lastPsmId,
Peptide = peptide,
Spectrum = spectrum,
Analysis = analysis,
ObservedNeutralMass = neutralPrecursorMass,
MonoisotopicMassError = neutralPrecursorMass - pwizPeptide.monoisotopicMass(),
MolecularWeightError = neutralPrecursorMass - pwizPeptide.molecularWeight(),
Charge = peptideTuple.Charge,
Rank = (peptideTuple.ScoreDivider == lastDivider ? rank : ++rank),
QValue = (rank == 1 ? row.QValue : PeptideSpectrumMatch.DefaultQValue),
};
if (row.Score != null)
psm.Scores = new Dictionary<string, double>()
{
{"score1", (double) row.Score/peptideTuple.ScoreDivider},
{"score2", 1/((double) row.Score/peptideTuple.ScoreDivider)}
};
bulkInserter.Add(psm);
lastDivider = peptideTuple.ScoreDivider;
// add PeptideModifications and Modifications
foreach (KeyValuePair<int, ModList> itr in pwizPeptide.modifications())
{
foreach (PwizMod pwizMod in itr.Value)
{
Modification mod = session.UniqueResult<Modification>(o => o.Formula == pwizMod.formula());
if (mod == null)
{
mod = new Modification()
{
Id = ++lastModId,
Formula = pwizMod.formula(),
MonoMassDelta = pwizMod.monoisotopicDeltaMass(),
AvgMassDelta = pwizMod.averageDeltaMass(),
Name = pwizMod.formula()
};
bulkInserter.Add(mod);
}
bulkInserter.Add(new PeptideModification()
{
Id = ++lastPmId,
PeptideSpectrumMatch = psm,
Modification = mod,
Offset = itr.Key == ModMap.NTerminus() ? int.MinValue
: itr.Key == ModMap.CTerminus() ? int.MaxValue
: itr.Key
});
}
}
}
}
}
bulkInserter.Execute();
bulkInserter.Reset("");
}
public void Add (PeptideSpectrumMatch psm)
{
insertRow(Table.PeptideSpectrumMatch, new object[]
{
psm.Id, psm.Peptide.Id, psm.Spectrum.Id, psm.Analysis.Id,
psm.ObservedNeutralMass,
psm.MonoisotopicMassError, psm.MolecularWeightError,
psm.Rank, psm.QValue, psm.Charge
});
if (psm.Scores != null)
foreach (var score in psm.Scores)
{
long scoreId;
if (!scoreIdByName.TryGetValue(score.Key, out scoreId))
{
if (scoreIdByName.Count > 0)
scoreIdByName[score.Key] = scoreId = scoreIdByName.Max(o => o.Value) + 1;
else
scoreIdByName[score.Key] = scoreId = 1;
insertRow(Table.PeptideSpectrumMatchScoreName, new object[] { scoreId, score.Key });
}
insertRow(Table.PeptideSpectrumMatchScore, new object[] {psm.Id, scoreId, score.Value});
}
}
private void SetSpectra(PhosphoRSInput input)
{
_spectra = input.Spectra;
_spectrumMap = new Dictionary <int, Spectrum>();
_peptideMap = new Dictionary <int, Peptide>();
_phosphoSiteMap = new Dictionary <int, List <int> >();
_items = new List <Tuple <PeptideSpectrumMatch, List <Tuple <int, List <int> > > > >();
var count = 0;
foreach (var spectrum in input.Spectra)
{
var rank = 0;
foreach (var peptide in spectrum.IdentifiedPhosphorPeptides)
{
if (_options.MaxHitRank > 0 && rank >= _options.MaxHitRank)
{
break;
}
rank += 1;
var isoformMap = new List <Tuple <int, List <int> > >();
var aas = AminoAcidSequence.Create(
peptide.ID,
peptide.Sequence,
_aminoAcidModifications,
peptide.Modification
);
var psm = new PeptideSpectrumMatch(
peptide.ID,
spectrum.ActivationType,
spectrum.PrecursorCharge,
1,
spectrum.Peaks.ToArray(),
aas
);
var phosphoCount = peptide.Modification
.Skip(2).SkipLast(2)
.Where(c => c == _options.PhosphorylationSymbol)
.Count();
if (phosphoCount == 0)
{
continue;
}
var targetSites = peptide.Sequence
.Select((c, i) => _phosphoTargetAminoAcids.Contains(c) ? i : -1)
.Where(i => i >= 0)
.ToList();
targetSites.Combinations(phosphoCount).Take(_options.MaxIsoformCount).ForEach(sites =>
{
var list = sites.ToList();
isoformMap.Add(
new Tuple <int, List <int> >(
count,
list
)
);
_spectrumMap.Add(count, spectrum);
_peptideMap.Add(count, peptide);
_phosphoSiteMap.Add(count, list);
count += 1;
});
_items.Add(Tuple.Create <PeptideSpectrumMatch, List <Tuple <int, List <int> > > >(
psm,
isoformMap
));
}
}
}