public static void CreateTestProteins(ISession session, IList<string> testProteinSequences)
{
var bulkInserter = new BulkInserter(session.Connection);
for (int i = 0; i < testProteinSequences.Count; ++i)
{
int id = i + 1;
var protein = new TestProtein()
{
Id = id,
Accession = "PRO" + id.ToString(),
Description = "Protein " + id.ToString(),
Sequence = testProteinSequences[i]
};
var proteinAccessor = new PrivateObject(protein);
proteinAccessor.SetProperty("Length", testProteinSequences[i].Length);
bulkInserter.Add(protein);
}
bulkInserter.Execute();
bulkInserter.Reset("");
}
private static void createTestPeptideInstances (NHibernate.ISession session, BulkInserter bulkInserter, Peptide pep)
{
// store instances even though the association is inverse:
// the PeptideModification.Offset property needs access to the protein sequence
pep.Instances = new List<PeptideInstance>();
foreach (Protein pro in session.Query<Protein>())
{
int start = pro.Sequence.IndexOf(pep.Sequence, 0);
while (start >= 0)
{
int end = start + pep.Sequence.Length;
bool nTerminusIsSpecific = start == 0 || pro.Sequence[start - 1] == 'K' || pro.Sequence[start - 1] == 'R';
bool cTerminusIsSpecific = end == pro.Sequence.Length || pro.Sequence[end - 1] == 'K' || pro.Sequence[end - 1] == 'R';
var instance = new TestPI()
{
Peptide = pep,
Protein = pro,
Offset = start,
Length = pep.Sequence.Length,
MissedCleavages = pep.Sequence.ToCharArray(0, pep.Sequence.Length - 1).Count(o => o == 'K' || o == 'R'),
NTerminusIsSpecific = nTerminusIsSpecific,
CTerminusIsSpecific = cTerminusIsSpecific,
};
var instanceAccessor = new PrivateObject(instance);
instanceAccessor.SetProperty("SpecificTermini", (nTerminusIsSpecific ? 1 : 0) + (cTerminusIsSpecific ? 1 : 0));
bulkInserter.Add(instance);
pep.Instances.Add(instance);
start = pro.Sequence.IndexOf(pep.Sequence, start + 1);
}
}
if (pep.Instances.Count == 0)
throw new ArgumentException("peptide " + pep.Sequence + " does not occur in any proteins");
}
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("");
}