public void TestFilter()
{
var pro1 = new IdentifiedProtein("P1");
pro1.Peptides.Add(new IdentifiedPeptide(new IdentifiedSpectrum() { Charge = 1 }));
pro1.Peptides.Add(new IdentifiedPeptide(new IdentifiedSpectrum() { Charge = 2 }));
var pro2 = new IdentifiedProtein("P2");
pro2.Peptides.Add(new IdentifiedPeptide(new IdentifiedSpectrum() { Charge = 3 }));
var g1 = new IdentifiedProteinGroup();
g1.Add(pro1);
var g2 = new IdentifiedProteinGroup();
g2.Add(pro2);
IdentifiedResult ir = new IdentifiedResult();
ir.Add(g1);
ir.Add(g2);
Assert.AreEqual(2, ir.Count);
Assert.AreEqual(3, ir.GetSpectra().Count);
ir.Filter(m => { return m.Spectrum.Query.Charge > 1; });
Assert.AreEqual(2, ir.Count);
Assert.AreEqual(2, ir.GetSpectra().Count);
ir.GetSpectra().All(m => { return m.Charge > 1; });
ir.Filter(m => { return m.Spectrum.Query.Charge > 2; });
Assert.AreEqual(1, ir.Count);
Assert.AreEqual(1, ir.GetSpectra().Count);
ir.GetSpectra().All(m => { return m.Charge > 2; });
Assert.AreEqual("P2", ir[0][0].Name);
}
public static List <IIdentifiedProtein> BuildProteins(IEnumerable <IIdentifiedSpectrum> peptides)
{
var proteinMap = new Dictionary <string, IIdentifiedProtein>();
foreach (IIdentifiedSpectrum mph in peptides)
{
foreach (IIdentifiedPeptide pep in mph.Peptides)
{
foreach (string proteinName in pep.Proteins)
{
if (!proteinMap.ContainsKey(proteinName))
{
IIdentifiedProtein protein = new IdentifiedProtein();
protein.Name = proteinName;
proteinMap[proteinName] = protein;
protein.Peptides.Add(pep);
}
else
{
proteinMap[proteinName].Peptides.Add(pep);
}
}
}
}
return(new List <IIdentifiedProtein>(proteinMap.Values));
}
public Dictionary <int, IIdentifiedProtein> ParseProteinMap(string fileName)
{
SQLiteDBHelper sqlite = new SQLiteDBHelper(fileName);
var result = new Dictionary <int, IIdentifiedProtein>();
string sqlProtein = "select ps.ProteinID, pa.Description, pro.Sequence, ps.ProteinScore, ps.Coverage from ProteinAnnotations as pa, Proteins as pro, ProteinScores as ps where pro.ProteinID=pa.ProteinID and pro.ProteinID=ps.ProteinID";
var proteinReader = sqlite.ExecuteReader(sqlProtein, null);
Progress.SetMessage("Parsing proteins ...");
while (proteinReader.Read())
{
var protein = new IdentifiedProtein();
var proid = proteinReader.GetInt32(0);
var des = proteinReader.GetString(1);
if (des.Length > 0 && des[0] == '>')
{
des = des.Substring(1);
}
protein.Reference = des;
protein.Sequence = proteinReader.GetString(2);
protein.Score = proteinReader.GetDouble(3);
protein.Coverage = proteinReader.GetDouble(4);
result[proid] = protein;
}
return(result);
}
private static List<IIdentifiedProtein> InitProteins()
{
var mph1 = new IdentifiedSpectrum();
mph1.Query.FileScan.Experimental = "EXP1";
var mp1 = new IdentifiedPeptide(mph1);
mp1.AddProtein("Protein1");
mp1.AddProtein("Protein2");
mp1.Sequence = "SEQ1";
var mph2 = new IdentifiedSpectrum();
mph2.Query.FileScan.Experimental = "EXP2";
var mp2 = new IdentifiedPeptide(mph2);
mp2.AddProtein("Protein1");
mp2.AddProtein("Protein3");
mp2.Sequence = "SEQ2";
var mpro1 = new IdentifiedProtein("Protein1");
mpro1.Peptides.Add(mp1);
mpro1.Peptides.Add(mp2);
var mpro2 = new IdentifiedProtein("Protein2");
mpro2.Peptides.Add(mp1);
var mpro3 = new IdentifiedProtein("Protein3");
mpro3.Peptides.Add(mp2);
var result = new List<IIdentifiedProtein>();
result.Add(mpro3);
result.Add(mpro2);
result.Add(mpro1);
return result;
}
public void TestSort()
{
var mph1 = new IdentifiedSpectrum();
var peptide1 = new IdentifiedPeptide(mph1);
var mph2 = new IdentifiedSpectrum();
var peptide2 = new IdentifiedPeptide(mph2);
var mpro1 = new IdentifiedProtein("Protein1");
mpro1.Peptides.Add(peptide1);
mpro1.Peptides.Add(peptide2);
var mpro2 = new IdentifiedProtein("Protein2");
mpro2.Peptides.Add(peptide1);
var mpro3 = new IdentifiedProtein("Protein3");
mpro3.Peptides.Add(peptide2);
var mpros = new List <IdentifiedProtein>();
mpros.Add(mpro3);
mpros.Add(mpro2);
mpros.Add(mpro1);
mpros.Sort();
Assert.AreEqual(mpro1, mpros[0]);
Assert.AreEqual(mpro2, mpros[1]);
Assert.AreEqual(mpro3, mpros[2]);
}
public void TestSort()
{
var mph1 = new IdentifiedSpectrum();
var peptide1 = new IdentifiedPeptide(mph1);
var mph2 = new IdentifiedSpectrum();
var peptide2 = new IdentifiedPeptide(mph2);
var mpro1 = new IdentifiedProtein("Protein1");
mpro1.Peptides.Add(peptide1);
mpro1.Peptides.Add(peptide2);
var mpro2 = new IdentifiedProtein("Protein2");
mpro2.Peptides.Add(peptide1);
var mpro3 = new IdentifiedProtein("Protein3");
mpro3.Peptides.Add(peptide2);
var mpros = new List<IdentifiedProtein>();
mpros.Add(mpro3);
mpros.Add(mpro2);
mpros.Add(mpro1);
mpros.Sort();
Assert.AreEqual(mpro1, mpros[0]);
Assert.AreEqual(mpro2, mpros[1]);
Assert.AreEqual(mpro3, mpros[2]);
}
public void TestKeepDistinctPeptideOnly()
{
var spectrum1 = new IdentifiedSpectrum();
var pep1 = spectrum1.NewPeptide();
var spectrum2 = new IdentifiedSpectrum();
var pep2 = spectrum2.NewPeptide();
var spectrum3 = new IdentifiedSpectrum();
var pep3 = spectrum3.NewPeptide();
var spectrum4 = new IdentifiedSpectrum();
var pep4 = spectrum4.NewPeptide();
var protein1 = new IdentifiedProtein();
protein1.Peptides.Add(pep1);
protein1.Peptides.Add(pep2);
var protein2 = new IdentifiedProtein();
protein2.Peptides.Add(pep1);
protein2.Peptides.Add(pep3);
var protein3 = new IdentifiedProtein();
protein3.Peptides.Add(pep2);
protein3.Peptides.Add(pep4);
var g1 = new IdentifiedProteinGroup();
g1.Add(protein1);
var g2 = new IdentifiedProteinGroup();
g2.Add(protein2);
var g3 = new IdentifiedProteinGroup();
g3.Add(protein3);
IIdentifiedResult ir = new IdentifiedResult();
ir.Add(g1);
ir.Add(g2);
ir.Add(g3);
new DistinctResultDistiller().KeepDistinctPeptideOnly(ir);
Assert.AreEqual(2, ir.Count);
Assert.AreEqual(g2, ir[0]);
Assert.AreEqual(g3, ir[1]);
Assert.AreEqual(1, ir[0].GetPeptides().Count);
Assert.AreEqual(spectrum3, ir[0].GetPeptides()[0]);
Assert.AreEqual(1, ir[1].GetPeptides().Count);
Assert.AreEqual(spectrum4, ir[1].GetPeptides()[0]);
}
public IIdentifiedProtein ParseString(string line)
{
IIdentifiedProtein result = new IdentifiedProtein();
this.converter.SetProperty(result, line);
return(result);
}
private IIdentifiedProteinGroup ReadNextProteinGroup(StreamReader filein,
Dictionary <string, IIdentifiedSpectrum> peptideMap,
ref string lastLine)
{
while (!IsProteinLine(lastLine) && (lastLine = filein.ReadLine()) != null)
{
}
if (lastLine == null)
{
return(null);
}
IIdentifiedProteinGroup result = new IdentifiedProteinGroup();
while (IsProteinLine(lastLine))
{
IIdentifiedProtein protein = new IdentifiedProtein();
this.proteinConverter.SetProperty(protein, lastLine);
result.Add(protein);
lastLine = filein.ReadLine();
}
var peptides = new List <IIdentifiedSpectrum>();
while (!IsProteinLine(lastLine))
{
IIdentifiedSpectrum mphit = new IdentifiedSpectrum();
this.peptideConverter.SetProperty(mphit, lastLine);
string id = mphit.Query.FileScan.LongFileName + "-" + mphit.Rank;
if (!peptideMap.ContainsKey(id))
{
peptideMap[id] = mphit;
}
else
{
mphit = peptideMap[id];
}
peptides.Add(mphit);
lastLine = filein.ReadLine();
if (lastLine == null || lastLine.Trim().Length == 0)
{
break;
}
}
peptides.Sort();
result.AddIdentifiedSpectra(peptides);
return(result);
}
private IdentifiedProtein GetProtein(String proteinContent)
{
List <String> proteinInfo = GetProteinInfo(proteinContent);
var result = new IdentifiedProtein();
result.Name = proteinInfo[0];
result.MolecularWeight = MyConvert.ToDouble(proteinInfo[1]);
result.Description = proteinInfo[2];
return(result);
}
public void TestReadWrite()
{
var ann = new IdentifiedProtein();
var pqr = new ProteinQuantificationResult();
pqr.Items["D1"] = new QuantificationItem()
{
Enabled = true,
Ratio = 1.5,
SampleIntensity = 150,
ReferenceIntensity = 100,
Correlation = 0.9,
ScanCount = 55,
Filename = "test1.silac"
};
pqr.Items["D2"] = new QuantificationItem()
{
Enabled = false,
Ratio = 3,
SampleIntensity = 250,
ReferenceIntensity = 200,
Correlation = 0.8,
ScanCount = 77,
Filename = "test2.silac"
};
ann.Annotations[SilacQuantificationConstants.SILAC_KEY] = pqr;
var converter = new SilacProteinQuantificationResultConverter2 <IAnnotation>();
List <IPropertyConverter <IAnnotation> > converters = new List <IPropertyConverter <IAnnotation> >();
converters.Add(converter);
converters.AddRange(converter.GetRelativeConverter(new IAnnotation[] { ann }.ToList()));
CompositePropertyConverter <IAnnotation> finalConverter = new CompositePropertyConverter <IAnnotation>(converters, ',');
Assert.AreEqual("S_COUNT,SE_D1,SR_D1,SRC_D1,SSI_D1,SRI_D1,SE_D2,SR_D2,SRC_D2,SSI_D2,SRI_D2", finalConverter.Name);
var line1 = finalConverter.GetProperty(ann);
Assert.AreEqual("2,True,1.5000,0.9000,150.0,100.0,False,3.0000,0.8000,250.0,200.0", line1);
var protein2 = new IdentifiedProtein();
var finalC = IdentifiedProteinPropertyConverterFactory.GetInstance().GetConverters(finalConverter.Name, ',');
finalC.SetProperty(protein2, line1);
var line2 = finalConverter.GetProperty(protein2);
Assert.AreEqual(line1, line2);
}
public void TestKeepDistinctPeptideOnly()
{
var spectrum1 = new IdentifiedSpectrum();
var pep1 = spectrum1.NewPeptide();
var spectrum2 = new IdentifiedSpectrum();
var pep2 = spectrum2.NewPeptide();
var spectrum3 = new IdentifiedSpectrum();
var pep3 = spectrum3.NewPeptide();
var spectrum4 = new IdentifiedSpectrum();
var pep4 = spectrum4.NewPeptide();
var protein1 = new IdentifiedProtein();
protein1.Peptides.Add(pep1);
protein1.Peptides.Add(pep2);
var protein2 = new IdentifiedProtein();
protein2.Peptides.Add(pep1);
protein2.Peptides.Add(pep3);
var protein3 = new IdentifiedProtein();
protein3.Peptides.Add(pep2);
protein3.Peptides.Add(pep4);
var g1 = new IdentifiedProteinGroup();
g1.Add(protein1);
var g2 = new IdentifiedProteinGroup();
g2.Add(protein2);
var g3 = new IdentifiedProteinGroup();
g3.Add(protein3);
IIdentifiedResult ir = new IdentifiedResult();
ir.Add(g1);
ir.Add(g2);
ir.Add(g3);
new DistinctResultDistiller().KeepDistinctPeptideOnly(ir);
Assert.AreEqual(2, ir.Count);
Assert.AreEqual(g2, ir[0]);
Assert.AreEqual(g3, ir[1]);
Assert.AreEqual(1, ir[0].GetPeptides().Count);
Assert.AreEqual(spectrum3, ir[0].GetPeptides()[0]);
Assert.AreEqual(1, ir[1].GetPeptides().Count);
Assert.AreEqual(spectrum4, ir[1].GetPeptides()[0]);
}
public void TestDistinctPeptides()
{
IdentifiedProtein protein = new IdentifiedProtein();
IdentifiedSpectrum sp1 = new IdentifiedSpectrum();
IdentifiedSpectrum sp2 = new IdentifiedSpectrum();
protein.Peptides.Add(new IdentifiedPeptide(sp1));
protein.Peptides.Add(new IdentifiedPeptide(sp1));
protein.Peptides.Add(new IdentifiedPeptide(sp2));
Assert.AreEqual(3, protein.Peptides.Count);
Assert.AreEqual(2, protein.GetSpectra().Count);
Assert.AreEqual(2, protein.GetDistinctPeptides().Count());
}
public void TestReadWrite()
{
var ann = new IdentifiedProtein();
var pqr = new ProteinQuantificationResult();
pqr.Items["D1"] = new QuantificationItem()
{
Enabled = true,
Ratio = 1.5,
SampleIntensity = 150,
ReferenceIntensity = 100,
Correlation = 0.9,
ScanCount = 55,
Filename = "test1.silac"
};
pqr.Items["D2"] = new QuantificationItem()
{
Enabled = false,
Ratio = 3,
SampleIntensity = 250,
ReferenceIntensity = 200,
Correlation = 0.8,
ScanCount = 77,
Filename = "test2.silac"
};
ann.Annotations[SilacQuantificationConstants.SILAC_KEY] = pqr;
var converter = new SilacProteinQuantificationResultConverter2<IAnnotation>();
List<IPropertyConverter<IAnnotation>> converters = new List<IPropertyConverter<IAnnotation>>();
converters.Add(converter);
converters.AddRange(converter.GetRelativeConverter(new IAnnotation[] { ann }.ToList()));
CompositePropertyConverter<IAnnotation> finalConverter = new CompositePropertyConverter<IAnnotation>(converters, ',');
Assert.AreEqual("S_COUNT,SE_D1,SR_D1,SRC_D1,SSI_D1,SRI_D1,SE_D2,SR_D2,SRC_D2,SSI_D2,SRI_D2", finalConverter.Name);
var line1 = finalConverter.GetProperty(ann);
Assert.AreEqual("2,True,1.5000,0.9000,150.0,100.0,False,3.0000,0.8000,250.0,200.0", line1);
var protein2 = new IdentifiedProtein();
var finalC = IdentifiedProteinPropertyConverterFactory.GetInstance().GetConverters(finalConverter.Name, ',');
finalC.SetProperty(protein2, line1);
var line2 = finalConverter.GetProperty(protein2);
Assert.AreEqual(line1, line2);
}
private static List <IIdentifiedProtein> InitProteins()
{
var mph1 = new IdentifiedSpectrum();
mph1.Query.FileScan.Experimental = "EXP1";
var mp1 = new IdentifiedPeptide(mph1);
mp1.AddProtein("Protein1");
mp1.AddProtein("Protein2");
mp1.Sequence = "SEQ1";
var mph2 = new IdentifiedSpectrum();
mph2.Query.FileScan.Experimental = "EXP2";
var mp2 = new IdentifiedPeptide(mph2);
mp2.AddProtein("Protein1");
mp2.AddProtein("Protein3");
mp2.Sequence = "SEQ2";
var mpro1 = new IdentifiedProtein("Protein1");
mpro1.Peptides.Add(mp1);
mpro1.Peptides.Add(mp2);
var mpro2 = new IdentifiedProtein("Protein2");
mpro2.Peptides.Add(mp1);
var mpro3 = new IdentifiedProtein("Protein3");
mpro3.Peptides.Add(mp2);
var result = new List <IIdentifiedProtein>();
result.Add(mpro3);
result.Add(mpro2);
result.Add(mpro1);
return(result);
}
protected List <IdentifiedProtein> ParseSameMatchProteins(String redundant)
{
var result = new List <IdentifiedProtein>();
Match tableMatch = GetTableRegex().Match(redundant);
while (tableMatch.Success)
{
try
{
IdentifiedProtein mp = GetProtein(tableMatch.Groups[1].Value);
result.Add(mp);
}
catch (ArgumentException)
{
}
tableMatch = tableMatch.NextMatch();
}
return(result);
}
public void TestFilter2()
{
var spectrum = new IdentifiedSpectrum();
spectrum.Query.FileScan.LongFileName = "ABDCDD.12.123.2.dat";
var pro1 = new IdentifiedProtein("P1");
pro1.Peptides.Add(new IdentifiedPeptide(spectrum) { Sequence = "AAAAAAA" });
var pro2 = new IdentifiedProtein("P2");
pro2.Peptides.Add(new IdentifiedPeptide(spectrum) { Sequence = "BBBBBBB" });
var g1 = new IdentifiedProteinGroup();
g1.Add(pro1);
g1.Add(pro2);
IdentifiedResult ir = new IdentifiedResult();
ir.Add(g1);
Assert.AreEqual(1, ir.Count);
Assert.AreEqual(2, ir[0].Count);
Assert.AreEqual(1, ir.GetSpectra().Count);
ir.Filter(m =>
{
return m.Sequence.Contains('A');
});
Assert.AreEqual(1, ir.Count);
Assert.AreEqual(1, ir[0].Count);
Assert.AreEqual(1, ir.GetSpectra().Count);
Assert.AreSame(pro1, ir[0][0]);
ir.Filter(m =>
{
return m.Sequence.Contains('C');
});
Assert.AreEqual(0, ir.Count);
}
public void TestGetAnnotationKeys()
{
string key1 = "TEST1";
string key2 = "TEST2";
var mph1 = new IdentifiedSpectrum();
mph1.Annotations.Add(key1, null);
mph1.Query.FileScan.Experimental = "EXP1";
new IdentifiedPeptide(mph1);
var mph2 = new IdentifiedSpectrum();
mph2.Annotations.Add(key2, null);
mph2.Query.FileScan.Experimental = "EXP2";
new IdentifiedPeptide(mph2);
mph1.Peptide.Sequence = "SEQ1";
mph2.Peptide.Sequence = "SEQ2";
var protein = new IdentifiedProtein();
protein.Peptides.Add(mph1.Peptide);
protein.Peptides.Add(mph2.Peptide);
var mpg = new IdentifiedProteinGroup();
mpg.Add(protein);
var mr = new MascotResult();
mr.Add(mpg);
List <string> annotationKeys = AnnotationUtils.GetAnnotationKeys(mr.GetSpectra());
Assert.AreEqual(2, annotationKeys.Count);
Assert.IsTrue(annotationKeys.Contains(key1));
Assert.IsTrue(annotationKeys.Contains(key2));
}
public void TestCalculateCoverage()
{
IdentifiedProtein protein = new IdentifiedProtein();
//total 30 amino acids
protein.Sequence = "ABCDEDFDEFDSESLKJFDJLSLGFGDDGD";
IdentifiedSpectrum s1 = new IdentifiedSpectrum();
IdentifiedPeptide p1 = new IdentifiedPeptide(s1);
p1.Sequence = "B.CDEDF.D";
protein.Peptides.Add(p1);
protein.CalculateCoverage();
Assert.AreEqual(16.67, protein.Coverage, 0.01);
IdentifiedSpectrum s2 = new IdentifiedSpectrum();
IdentifiedPeptide p2 = new IdentifiedPeptide(s2);
p2.Sequence = "F.DSESL.K";
protein.Peptides.Add(p2);
protein.CalculateCoverage();
Assert.AreEqual(33.33, protein.Coverage, 0.01);
IdentifiedSpectrum s3 = new IdentifiedSpectrum();
IdentifiedPeptide p3 = new IdentifiedPeptide(s3);
p3.Sequence = "L.SLGF.G";
protein.Peptides.Add(p3);
protein.CalculateCoverage();
Assert.AreEqual(46.67, protein.Coverage, 0.01);
IdentifiedSpectrum s4 = new IdentifiedSpectrum();
IdentifiedPeptide p4 = new IdentifiedPeptide(s4);
p4.Sequence = "L.SLGFG.D";
protein.Peptides.Add(p4);
protein.CalculateCoverage();
Assert.AreEqual(50.00, protein.Coverage, 0.01);
}
public Dictionary <int, IIdentifiedProtein> ParseProteinMap(string fileName, bool isDecoy)
{
var suffix = isDecoy ? "_decoy" : "";
SQLiteDBHelper sqlite = new SQLiteDBHelper(fileName);
var result = new Dictionary <int, IIdentifiedProtein>();
string sqlProtein = string.Format("select ps.ProteinID, pa.Description, pro.Sequence, ps.ProteinScore, ps.Coverage from ProteinAnnotations as pa, Proteins as pro, ProteinScores{0} as ps where pro.ProteinID=pa.ProteinID and pro.ProteinID=ps.ProteinID", suffix);
var proteinReader = sqlite.ExecuteReader(sqlProtein, null);
Progress.SetMessage("Parsing proteins ...");
while (proteinReader.Read())
{
var protein = new IdentifiedProtein();
var proid = proteinReader.GetInt32(0);
var des = proteinReader.GetString(1);
if (des.Length > 0 && des[0] == '>')
{
des = des.Substring(1);
}
protein.Reference = des;
protein.Sequence = proteinReader.GetString(2);
protein.Score = proteinReader.GetDouble(3);
protein.Coverage = proteinReader.GetDouble(4);
result[proid] = protein;
}
if (isDecoy)
{
foreach (var v in result.Values)
{
v.Sequence = SequenceUtils.GetReversedSequence(v.Sequence);
v.Reference = GetReversedReference(v.Reference);
}
}
return(result);
}
public void TestDtaselect()
{
string header = "Locus Sequence Count Spectrum Count Sequence Coverage Length MolWt pI Validation Status Descriptive Name";
IPropertyConverter<IIdentifiedProtein> converter = IdentifiedProteinPropertyConverterFactory.GetInstance().GetConverters(header, '\t');
Assert.AreEqual(header, converter.Name);
string line = "YDR050C 495 495 81.10% 249 26795.41 5.74 U YDR050C TPI1 SGDID:S000002457, Chr IV from 556470-555724, reverse complement, Verified ORF, \"Triose phosphate isomerase, abundant glycolytic enzyme; mRNA half-life is regulated by iron availability; transcription is controlled by activators Reb1p, Gcr1p, and Rap1p through binding sites in the 5' non-coding region\"";
IIdentifiedProtein protein = new IdentifiedProtein();
converter.SetProperty(protein, line);
Assert.AreEqual("YDR050C", protein.Name);
Assert.AreEqual("TPI1 SGDID:S000002457, Chr IV from 556470-555724, reverse complement, Verified ORF, \"Triose phosphate isomerase, abundant glycolytic enzyme; mRNA half-life is regulated by iron availability; transcription is controlled by activators Reb1p, Gcr1p, and Rap1p through binding sites in the 5' non-coding region\"", protein.Description);
Assert.AreEqual(495, protein.UniquePeptideCount);
Assert.AreEqual(81.1, protein.Coverage);
Assert.AreEqual(26795.41, protein.MolecularWeight);
Assert.AreEqual(5.74, protein.IsoelectricPoint);
for (int i = 0; i < 495; i++)
{
protein.Peptides.Add(new IdentifiedPeptide( new IdentifiedSpectrum()));
}
Assert.AreEqual(line, converter.GetProperty(protein));
}
public void TestUniqueCount()
{
var protein = new IdentifiedProtein();
var spectrum = new IdentifiedSpectrum();
var peptides = new List<IIdentifiedPeptide>();
peptides.Add(new IdentifiedPeptide(spectrum){
Sequence = "ILLLAR"
});
peptides.Add(new IdentifiedPeptide(spectrum){
Sequence = "LILIAR"
});
Assert.AreEqual(1, IdentifiedPeptideUtils.GetUniquePeptideCount(peptides));
peptides.Add(new IdentifiedPeptide(new IdentifiedSpectrum())
{
Sequence = "LIIIAR"
});
Assert.AreEqual(1, IdentifiedPeptideUtils.GetUniquePeptideCount(peptides));
}
public List <Sequence> ParseProteinSequences(string fileName)
{
SQLiteDBHelper sqlite = new SQLiteDBHelper(fileName);
var result = new List <Sequence>();
string sqlProtein = "select pa.Description, pro.Sequence from ProteinAnnotations as pa, Proteins as pro where pro.ProteinID=pa.ProteinID";
var proteinReader = sqlite.ExecuteReader(sqlProtein, null);
Progress.SetMessage("Parsing proteins ...");
while (proteinReader.Read())
{
var protein = new IdentifiedProtein();
var des = proteinReader.GetString(0);
if (des.Length > 0 && des[0] == '>')
{
des = des.Substring(1);
}
result.Add(new Sequence(des, proteinReader.GetString(1)));
}
return(result);
}
public void TestBuild()
{
var pep1 = new IdentifiedPeptide(new IdentifiedSpectrum(new SequestFilename("A", 1, 1, 1, ".dta"))) { Sequence = "A" };
var pep2 = new IdentifiedPeptide(new IdentifiedSpectrum(new SequestFilename("B", 1, 1, 1, ".dta"))) { Sequence = "B" };
var pep3 = new IdentifiedPeptide(new IdentifiedSpectrum(new SequestFilename("C", 1, 1, 1, ".dta"))) { Sequence = "C" };
var pep4 = new IdentifiedPeptide(new IdentifiedSpectrum(new SequestFilename("D", 1, 1, 1, ".dta"))) { Sequence = "D" };
var pep5 = new IdentifiedPeptide(new IdentifiedSpectrum(new SequestFilename("E", 1, 1, 1, ".dta"))) { Sequence = "E" };
var pep6 = new IdentifiedPeptide(new IdentifiedSpectrum(new SequestFilename("F", 1, 1, 1, ".dta"))) { Sequence = "F" };
var protein1 = new IdentifiedProtein()
{
Peptides = new IIdentifiedPeptide[] { pep1, pep3, pep5, pep6 }.ToList()
};
var protein2 = new IdentifiedProtein()
{
Peptides = new IIdentifiedPeptide[] { pep2, pep3, pep4 }.ToList()
};
//should be removed from final result since all peptides has been included in protein1 and protein2, even one protein contains both peptides
var protein3 = new IdentifiedProtein()
{
Peptides = new IIdentifiedPeptide[] { pep1, pep2 }.ToList()
};
//should be removed from final result since all peptides has been included in protein1
var protein4 = new IdentifiedProtein()
{
Peptides = new IIdentifiedPeptide[] { pep1, pep5 }.ToList()
};
var actual = new IdentifiedProteinGroupBuilder().Build(new IIdentifiedProtein[] { protein1, protein2, protein3 }.ToList());
Assert.AreEqual(2, actual.Count);
Assert.AreSame(protein1, actual[0][0]);
Assert.AreSame(protein2, actual[1][0]);
}
public void TestNoredundant()
{
string header = " Reference PepCount UniquePepCount CoverPercent MW PI IdentifiedName";
IPropertyConverter<IIdentifiedProtein> converter = IdentifiedProteinPropertyConverterFactory.GetInstance().GetConverters(header, '\t');
Assert.AreEqual(header, converter.Name);
string line = "\tIPI:IPI00784154.1|SWISS-PROT:P10809|TREMBL:B2R5M6;Q53QD5;Q53SE2;Q96RI4;Q9UCR6|ENSEMBL:ENSP00000340019;ENSP00000373620|REFSEQ:NP_002147;NP_955472|H-INV:HIT000031088 Tax_Id=9606 Gene_Symbol=HSPD1 60 kDa heat shock protein, mitochondrial 84 19 43.46% 61054.43 5.70 IPI:IPI00784154.1|SWISS-PROT:P10809|TREMBL:B2R5M6;Q53QD5;Q53SE2;Q96RI4;Q9UCR6|ENSEMBL:ENSP00000340019;ENSP00000373620|REFSEQ:NP_002147;NP_955472|H-INV:HIT000031088 Tax_Id=9606 Gene_Symbol=HSPD1 60 kDa heat shock protein, mitochondrial";
IIdentifiedProtein protein = new IdentifiedProtein();
converter.SetProperty(protein, line);
Assert.AreEqual("IPI:IPI00784154.1|SWISS-PROT:P10809|TREMBL:B2R5M6;Q53QD5;Q53SE2;Q96RI4;Q9UCR6|ENSEMBL:ENSP00000340019;ENSP00000373620|REFSEQ:NP_002147;NP_955472|H-INV:HIT000031088", protein.Name);
Assert.AreEqual("Tax_Id=9606 Gene_Symbol=HSPD1 60 kDa heat shock protein, mitochondrial", protein.Description);
Assert.AreEqual(19, protein.UniquePeptideCount);
Assert.AreEqual(43.46, protein.Coverage);
Assert.AreEqual(61054.43, protein.MolecularWeight);
Assert.AreEqual(5.7, protein.IsoelectricPoint);
for (int i = 0; i < 84; i++)
{
protein.Peptides.Add(new IdentifiedPeptide( new IdentifiedSpectrum()));
}
Assert.AreEqual(line, converter.GetProperty(protein));
}
public bool IsProteinLine(string line)
{
if (null == line)
{
return(false);
}
if (line.StartsWith("\t"))
{
return(false);
}
try
{
var mp = new IdentifiedProtein();
this.proteinConverter.SetProperty(mp, line);
}
catch (Exception)
{
return(false);
}
return(true);
}
public void TestReadWrite()
{
var ann = new IdentifiedProtein();
var pqr = new ITraqQuantificationResult();
pqr.DatasetMap["D1"] = new ITraqQuantificationDatasetItem()
{
DatasetName = "D1"
};
pqr.DatasetMap["D1"].RatioMap["R114/REF"] = new ITraqQuantificationChannelItem()
{
ChannelName = "R114/REF",
Ratio = 1.5
};
pqr.DatasetMap["D1"].RatioMap["R115/REF"] = new ITraqQuantificationChannelItem()
{
ChannelName = "R115/REF",
Ratio = 1.8
};
pqr.DatasetMap["D2"] = new ITraqQuantificationDatasetItem()
{
DatasetName = "D2"
};
pqr.DatasetMap["D2"].RatioMap["R116/REF"] = new ITraqQuantificationChannelItem()
{
ChannelName = "R116/REF",
Ratio = 2.5
};
pqr.DatasetMap["D2"].RatioMap["R117/REF"] = new ITraqQuantificationChannelItem()
{
ChannelName = "R117/REF",
Ratio = 3.8
};
ann.Annotations[ITraqConsts.ITRAQ_KEY] = pqr;
var converter = new ITraqQuantificationResultConverter <IAnnotation>();
List <IPropertyConverter <IAnnotation> > converters = new List <IPropertyConverter <IAnnotation> >();
converters.Add(converter);
converters.AddRange(converter.GetRelativeConverter(new IAnnotation[] { ann }.ToList()));
CompositePropertyConverter <IAnnotation> finalConverter = new CompositePropertyConverter <IAnnotation>(converters, ',');
if (exportToConsole)
{
Console.WriteLine(finalConverter.Name);
}
Assert.AreEqual("ITRAQ_COUNT,IR_D1_R114/REF,IR_D1_R115/REF,IR_D2_R116/REF,IR_D2_R117/REF", finalConverter.Name);
var line1 = finalConverter.GetProperty(ann);
if (exportToConsole)
{
Console.WriteLine(line1);
}
Assert.AreEqual("2,1.5000,1.8000,2.5000,3.8000", line1);
var protein2 = new IdentifiedProtein();
var finalC = IdentifiedProteinPropertyConverterFactory.GetInstance().GetConverters(finalConverter.Name, ',');
finalC.SetProperty(protein2, line1);
var line2 = finalConverter.GetProperty(protein2);
Assert.AreEqual(line1, line2);
}
public MascotResult ParseContent(String fileContent)
{
var result = new MascotResult();
this.modifications = new Dictionary <string, char>();
Pair <int, double> pValueScore = ParsePValueScore(fileContent);
result.PValueScore = pValueScore.First;
result.PValue = pValueScore.Second;
var offsets = new List <Offset>();
try
{
result.PeakIsotopicType = ParsePeakIsotopicType(fileContent);
}
catch (ArgumentException)
{
}
try
{
result.PeakTolerance = ParsePeakTolerance(fileContent);
}
catch (ArgumentException)
{
}
var filters = new List <IFilter <IIdentifiedSpectrum> >();
if (this.filterByDefaultScoreAndPvalue)
{
filters.Add(new IdentifiedSpectrumScoreFilter(pValueScore.First));
filters.Add(new IdentifiedSpectrumExpectValueFilter(pValueScore.Second));
}
filters.Add(new IdentifiedSpectrumRankFilter(1));
if (null != this.defaultPeptideFilter)
{
filters.Add(this.defaultPeptideFilter);
}
this.currentPeptideFilter = new AndFilter <IIdentifiedSpectrum>(filters);
Match proteinMatch = GetProteinRegex().Match(fileContent);
while (proteinMatch.Success)
{
IdentifiedProtein protein = ParseProtein(proteinMatch.Groups[1].Value);
var group = new IdentifiedProteinGroup();
group.Add(protein);
result.Add(group);
offsets.Add(new Offset(proteinMatch.Index, proteinMatch.Index + proteinMatch.Length, group));
proteinMatch = proteinMatch.NextMatch();
}
int endIndex = fileContent.IndexOf("Peptide matches not assigned to protein hits");
if (-1 == endIndex)
{
endIndex = fileContent.Length - 1;
}
for (int i = 0; i < offsets.Count; i++)
{
int start = offsets[i].End;
int end = i == offsets.Count - 1 ? endIndex : offsets[i + 1].Start;
String redundant = fileContent.Substring(start, end - start + 1);
if (!redundant.Contains("Proteins matching the same set"))
{
continue;
}
List <IdentifiedProtein> sameMatchProteins = ParseSameMatchProteins(redundant);
foreach (IdentifiedProtein mp in sameMatchProteins)
{
mp.Peptides.AddRange(offsets[i].Mpg[0].Peptides);
offsets[i].Mpg.Add(mp);
}
}
for (int i = result.Count - 1; i >= 0; i--)
{
if (0 == result[i][0].Peptides.Count)
{
result.RemoveAt(i);
}
}
RefineModification(result);
MergePeptides(result);
result.InitUniquePeptideCount();
return(result);
}
public void TestGetAnnotationKeys()
{
string key1 = "TEST1";
string key2 = "TEST2";
var mph1 = new IdentifiedSpectrum();
mph1.Annotations.Add(key1, null);
mph1.Query.FileScan.Experimental = "EXP1";
new IdentifiedPeptide(mph1);
var mph2 = new IdentifiedSpectrum();
mph2.Annotations.Add(key2, null);
mph2.Query.FileScan.Experimental = "EXP2";
new IdentifiedPeptide(mph2);
mph1.Peptide.Sequence = "SEQ1";
mph2.Peptide.Sequence = "SEQ2";
var protein = new IdentifiedProtein();
protein.Peptides.Add(mph1.Peptide);
protein.Peptides.Add(mph2.Peptide);
var mpg = new IdentifiedProteinGroup();
mpg.Add(protein);
var mr = new MascotResult();
mr.Add(mpg);
List<string> annotationKeys = AnnotationUtils.GetAnnotationKeys(mr.GetSpectra());
Assert.AreEqual(2, annotationKeys.Count);
Assert.IsTrue(annotationKeys.Contains(key1));
Assert.IsTrue(annotationKeys.Contains(key2));
}
public override List <IIdentifiedProtein> ParseProteins(string fileName)
{
Dictionary <string, IIdentifiedProtein> proteinMap = new Dictionary <string, IIdentifiedProtein>();
using (StreamReader sr = new StreamReader(fileName))
{
string line = sr.ReadLine();
string[] headerParts = line.Split('\t');
int seqIndex = Array.FindIndex(headerParts, (m => m == "Sequence"));
int proIndex = Array.FindIndex(headerParts, (m => m == "Protein Accessions"));
int modIndex = Array.FindIndex(headerParts, (m => m == "Modifications"));
int xcIndex = Array.FindIndex(headerParts, (m => m == "XCorr"));
int deltaIndex = Array.FindIndex(headerParts, (m => m.EndsWith(" Score")));
int chargeIndex = Array.FindIndex(headerParts, (m => m == "Charge"));
int obsIndex = Array.FindIndex(headerParts, (m => m == "m/z [Da]"));
int mhIndex = Array.FindIndex(headerParts, (m => m == "MH+ [Da]"));
int fscanIndex = Array.FindIndex(headerParts, (m => m == "First Scan"));
int lscanIndex = Array.FindIndex(headerParts, (m => m == "Last Scan"));
int ionIndex = Array.FindIndex(headerParts, (m => m == "Ions Matched"));
int fileIndex = Array.FindIndex(headerParts, (m => m == "Spectrum File"));
Progress.SetRange(0, sr.BaseStream.Length);
Progress.SetMessage("Parsing file ...");
while ((line = sr.ReadLine()) != null)
{
if (line.Trim().Length == 0)
{
break;
}
string[] parts = line.Split('\t');
if (parts[0].Length == 0)
{
continue;
}
Progress.SetPosition(sr.BaseStream.Position);
string seq = parts[seqIndex];
string deltaCn = parts[deltaIndex];
if (deltaCn.Length == 0)//rank > 1
{
continue;
}
string protein = parts[proIndex];
if (!proteinMap.ContainsKey(protein))
{
sr.ReadLine();
string proLine = sr.ReadLine();
string[] proParts = proLine.Split('\t');
var p = new IdentifiedProtein(protein);
p.Coverage = MyConvert.ToDouble(proParts[2]);
p.MolecularWeight = MyConvert.ToDouble(proParts[5]) * 1000;
p.IsoelectricPoint = MyConvert.ToDouble(proParts[6]);
p.Score = MyConvert.ToDouble(proParts[7]);
p.Description = proParts[8];
proteinMap[protein] = p;
}
var pro = proteinMap[protein];
IdentifiedSpectrum spectrum = new IdentifiedSpectrum();
IdentifiedPeptide peptide = new IdentifiedPeptide(spectrum);
peptide.Sequence = seq.ToUpper();
peptide.AddProtein(protein);
spectrum.Modifications = parts[modIndex];
spectrum.DeltaScore = MyConvert.ToDouble(deltaCn);
spectrum.Charge = Convert.ToInt32(parts[chargeIndex]);
spectrum.ObservedMz = MyConvert.ToDouble(parts[obsIndex]);
spectrum.TheoreticalMH = MyConvert.ToDouble(parts[mhIndex]);
spectrum.Ions = parts[ionIndex];
spectrum.Query.FileScan.FirstScan = Convert.ToInt32(parts[fscanIndex]);
spectrum.Query.FileScan.LastScan = Convert.ToInt32(parts[lscanIndex]);
spectrum.Query.FileScan.Experimental = FileUtils.RemoveAllExtension(parts[fileIndex]);
pro.Peptides.Add(peptide);
}
}
var proteins = proteinMap.Values.ToList();
return(proteins);
}
public override List <IIdentifiedProtein> ParseProteins(string fileName)
{
Dictionary <string, IIdentifiedProtein> proteinMap = new Dictionary <string, IIdentifiedProtein>();
Application xApp = new Application();
//得到WorkBook对象, 可以用两种方式之一: 下面的是打开已有的文件
Workbook xBook = xApp.Workbooks._Open(fileName,
Missing.Value, Missing.Value, Missing.Value, Missing.Value
, Missing.Value, Missing.Value, Missing.Value, Missing.Value
, Missing.Value, Missing.Value, Missing.Value, Missing.Value);
try
{
Worksheet xSheet = (Worksheet)xBook.Sheets[1];
int fromRow = 2;
int endRow = fromRow;
for (; endRow <= xSheet.Rows.Count; endRow++)
{
string b = xSheet.Value('B', endRow);
if (null == b)
{
break;
}
}
endRow--;
Progress.SetRange(fromRow, endRow);
Progress.SetMessage("Parsing file ...");
for (int i = fromRow; i <= endRow; i++)
{
Progress.SetPosition(i);
string seq = xSheet.Value('A', i);
if (null == seq)//蛋白质信息
{
continue;
}
string deltaCn = xSheet.Value('I', i);
if (null == deltaCn)//rank > 1
{
continue;
}
string protein = xSheet.Value('B', i);
if (!proteinMap.ContainsKey(protein))
{
var p = new IdentifiedProtein(protein);
p.Coverage = MyConvert.ToDouble(xSheet.Value('C', i + 2));
p.MolecularWeight = MyConvert.ToDouble(xSheet.Value('F', i + 2)) * 1000;
p.IsoelectricPoint = MyConvert.ToDouble(xSheet.Value('G', i + 2));
p.Score = MyConvert.ToDouble(xSheet.Value('H', i + 2));
p.Description = xSheet.Value('I', i + 2);
proteinMap[protein] = p;
}
var pro = proteinMap[protein];
IdentifiedSpectrum spectrum = new IdentifiedSpectrum();
IdentifiedPeptide peptide = new IdentifiedPeptide(spectrum);
peptide.Sequence = seq.ToUpper();
peptide.AddProtein(protein);
spectrum.Modifications = xSheet.Value('F', i);
spectrum.DeltaScore = MyConvert.ToDouble(deltaCn);
spectrum.Charge = Convert.ToInt32(xSheet.Value('K', i));
spectrum.ObservedMz = MyConvert.ToDouble(xSheet.Value('L', i));
spectrum.TheoreticalMH = MyConvert.ToDouble(xSheet.Value('M', i));
spectrum.Ions = xSheet.Value('S', i);
spectrum.Query.FileScan.FirstScan = Convert.ToInt32(xSheet.Value('P', i));
spectrum.Query.FileScan.LastScan = Convert.ToInt32(xSheet.Value('Q', i));
spectrum.Query.FileScan.Experimental = FileUtils.RemoveAllExtension(xSheet.Value('T', i));
pro.Peptides.Add(peptide);
}
}
finally
{
xBook.Close(false, Type.Missing, Type.Missing);
}
var proteins = proteinMap.Values.ToList();
return(proteins);
}
protected IdentifiedProtein ParseProtein(String proteinContent)
{
IdentifiedProtein result = GetProtein(proteinContent);
List <String> peptideInfoContentList = GetPeptideInfoContentList(proteinContent);
foreach (String peptideInfoContent in peptideInfoContentList)
{
List <String> peptideInfo = GetPeptideInfo(peptideInfoContent);
if (0 == peptideInfo.Count)
{
continue;
}
IIdentifiedSpectrum mphit = new IdentifiedSpectrum();
// Group 0 : peptide mass from observed m/z
double experimentalPeptideMass = MyConvert.ToDouble(peptideInfo[0]);
mphit.ExperimentalMass = experimentalPeptideMass;
// Group 1 : observed m/z
double observed = MyConvert.ToDouble(peptideInfo[1]);
mphit.Query.ObservedMz = observed;
// Group 2 : charge
int charge = int.Parse(peptideInfo[2]);
mphit.Query.Charge = charge;
// Group 3 : title
String title = Uri.UnescapeDataString(peptideInfo[3]).Trim();
mphit.Query.Title = title;
SequestFilename sf = MascotUtils.ParseTitle(title, charge);
if (sf != null)
{
mphit.Query.FileScan.LongFileName = sf.LongFileName;
}
// Group 4 : query
mphit.Query.QueryId = int.Parse(peptideInfo[4]);
// Group 5 equals Group 1
// Group 6 equals Group 0
// Group 7 : calculated peptide mass
mphit.TheoreticalMass = MyConvert.ToDouble(peptideInfo[7]);
// Group 8 : different between observed peptide mass and calculated
// peptide mass
// Group 9 : miss cleavage
mphit.NumMissedCleavages = int.Parse(peptideInfo[9]);
// Group 10: score
mphit.Score = int.Parse(peptideInfo[10]);
// Group 11: expect p value
mphit.ExpectValue = MyConvert.ToDouble(peptideInfo[11]);
// Group 12: rank
mphit.Rank = int.Parse(peptideInfo[12]);
// Group 13: peptide sequence
// K.YEINVLR<u>.</u>N + Label:18O(2) (C-term)
String seq = peptideInfo[13].Replace(" ", "");
var mpep = new IdentifiedPeptide(mphit);
string[] parts = Regex.Split(seq, "\\+");
if (parts.Length > 1)
{
seq = parts[0].Trim();
mphit.Modifications = parts[1].Trim();
string[] mods = parts[1].Trim().Split(new[] { ';' }, StringSplitOptions.RemoveEmptyEntries);
foreach (string mod in mods)
{
Match m = this.modificationReg.Match(mod.Trim());
if (!this.modifications.ContainsKey(m.Groups[1].Value))
{
this.modifications[m.Groups[1].Value] = ' ';
}
}
}
mpep.Sequence = seq;
if (GetPeptideFilter().Accept(mphit))
{
mpep.AddProtein(result.Name);
result.Peptides.Add(mpep);
}
}
return(result);
}
public void TestReadWrite()
{
var ann = new IdentifiedProtein();
var pqr = new ITraqQuantificationResult();
pqr.DatasetMap["D1"] = new ITraqQuantificationDatasetItem()
{
DatasetName = "D1"
};
pqr.DatasetMap["D1"].RatioMap["R114/REF"] = new ITraqQuantificationChannelItem()
{
ChannelName = "R114/REF",
Ratio = 1.5
};
pqr.DatasetMap["D1"].RatioMap["R115/REF"] = new ITraqQuantificationChannelItem()
{
ChannelName = "R115/REF",
Ratio = 1.8
};
pqr.DatasetMap["D2"] = new ITraqQuantificationDatasetItem()
{
DatasetName = "D2"
};
pqr.DatasetMap["D2"].RatioMap["R116/REF"] = new ITraqQuantificationChannelItem()
{
ChannelName = "R116/REF",
Ratio = 2.5
};
pqr.DatasetMap["D2"].RatioMap["R117/REF"] = new ITraqQuantificationChannelItem()
{
ChannelName = "R117/REF",
Ratio = 3.8
};
ann.Annotations[ITraqConsts.ITRAQ_KEY] = pqr;
var converter = new ITraqQuantificationResultConverter<IAnnotation>();
List<IPropertyConverter<IAnnotation>> converters = new List<IPropertyConverter<IAnnotation>>();
converters.Add(converter);
converters.AddRange(converter.GetRelativeConverter(new IAnnotation[] { ann }.ToList()));
CompositePropertyConverter<IAnnotation> finalConverter = new CompositePropertyConverter<IAnnotation>(converters, ',');
if (exportToConsole)
{
Console.WriteLine(finalConverter.Name);
}
Assert.AreEqual("ITRAQ_COUNT,IR_D1_R114/REF,IR_D1_R115/REF,IR_D2_R116/REF,IR_D2_R117/REF", finalConverter.Name);
var line1 = finalConverter.GetProperty(ann);
if (exportToConsole)
{
Console.WriteLine(line1);
}
Assert.AreEqual("2,1.5000,1.8000,2.5000,3.8000", line1);
var protein2 = new IdentifiedProtein();
var finalC = IdentifiedProteinPropertyConverterFactory.GetInstance().GetConverters(finalConverter.Name, ',');
finalC.SetProperty(protein2, line1);
var line2 = finalConverter.GetProperty(protein2);
Assert.AreEqual(line1, line2);
}
