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 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 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 List <IIdentifiedProtein> Build <T>(IEnumerable <T> spectra) where T : IIdentifiedSpectrumBase
{
Dictionary <string, IIdentifiedProtein> proteins = new Dictionary <string, IIdentifiedProtein>();
HashSet <string> inserted = new HashSet <string>();
foreach (var spectrum in spectra)
{
inserted.Clear();
foreach (var peptide in spectrum.Peptides)
{
foreach (string ac in peptide.Proteins)
{
//如果一个蛋白中多个肽段都对应了这个谱图，只选择第一个肽段加入谱图
if (inserted.Contains(ac))
{
continue;
}
inserted.Add(ac);
if (!proteins.ContainsKey(ac))
{
IIdentifiedProtein pro = new IdentifiedProtein();
pro.Name = ac;
proteins[ac] = pro;
}
proteins[ac].Peptides.Add(peptide);
}
}
}
return(new List <IIdentifiedProtein>(proteins.Values));
}
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 static IIdentifiedResult DoBuildGroupByPeptide(List <IIdentifiedSpectrum> spectra, Func <IIdentifiedPeptide, string> func)
{
IdentifiedResult result = new IdentifiedResult();
var singlePeptides = (from s in spectra
where s.Peptides.Count == 1
select s).GroupBy(m => func(m.Peptide));
var multiplePeptides = (from s in spectra
where s.Peptides.Count > 1
select s).ToList();
Dictionary <string, List <IIdentifiedPeptide> > dic = new Dictionary <string, List <IIdentifiedPeptide> >();
foreach (var g in singlePeptides)
{
dic[g.Key] = new List <IIdentifiedPeptide>(from s in g
select s.Peptide);
}
foreach (var o in multiplePeptides)
{
var pc = (from p in o.Peptides
let c = dic.ContainsKey(func(p)) ? dic[func(p)].Count : 0
orderby c descending
select p).First();
if (!dic.ContainsKey(func(pc)))
{
dic[func(pc)] = new List <IIdentifiedPeptide>();
}
dic[func(pc)].Add(pc);
}
var keys = new List <string>(dic.Keys);
keys.Sort();
foreach (var key in keys)
{
IdentifiedProtein protein = new IdentifiedProtein(key);
protein.Peptides.AddRange(dic[key]);
protein.UniquePeptideCount = 1;
protein.Description = dic[key][0].Proteins.Merge('/');
IdentifiedProteinGroup group = new IdentifiedProteinGroup();
group.Add(protein);
result.Add(group);
}
result.BuildGroupIndex();
//result.Sort();
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 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 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 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);
}
private IdentifiedProtein ReadIndividualProtein(XmlTextReader reader)
{
IdentifiedProtein result = new IdentifiedProtein();
Dictionary <string, string> elements = new Dictionary <string, string>();
String nodeName = "";
while (reader.Read())
{
switch (reader.NodeType)
{
case XmlNodeType.Element:
if (reader.Name.Equals(peptideTag))
{
result.Peptides.Add(ReadIndividualPeptide(reader, result.Name).Peptide);
}
else
{
nodeName = reader.Name;
}
break;
case XmlNodeType.Text:
if (nodeName.Equals("reference"))
{
result.Reference = reader.Value;
}
else
{
elements.Add(nodeName, reader.Value);
}
break;
case XmlNodeType.EndElement:
if (reader.Name.Equals(proteinTag))
{
AssignProteinValue(result, elements);
return(result);
}
break;
}
}
return(result);
}
