public override void SetProperty(T t, string value)
{
t.ClearPeptides();
IIdentifiedPeptide mp = t.NewPeptide();
mp.Sequence = value;
}
public override IEnumerable <string> Process(string fileName)
{
MascotResultTextFormat format = new MascotResultTextFormat();
IIdentifiedResult ir = format.ReadFromFile(fileName);
List <IIdentifiedSpectrum> spectra = ir.GetSpectra();
spectra.ForEach(m =>
{
for (int i = m.Peptides.Count - 1; i >= 0; i--)
{
IIdentifiedPeptide peptide = m.Peptides[i];
string seq = PeptideUtils.GetMatchedSequence(peptide.Sequence);
if (!validator.Validate(seq))
{
m.RemovePeptideAt(i);
peptide.Spectrum = null;
}
}
});
ir.Filter(m =>
{
return(m.Spectrum != null);
});
string result = fileName + ".Labeled";
format.WriteToFile(result, ir);
return(new[] { result });
}
private static void WritePeptideProtein(StreamWriter sw, IIdentifiedPeptide pep, int iProtein)
{
sw.WriteLine(" <alternative_protein protein=\"{0}\" peptide_prev_aa=\"{1}\" peptide_next_aa=\"{2}\"/>",
pep.Proteins[iProtein],
pep.Sequence[1] == '.' ? pep.Sequence[0] : ' ',
pep.Sequence[pep.Sequence.Length - 2] == '.' ? pep.Sequence[pep.Sequence.Length - 1] : ' ');
}
private void MergePeptides(MascotResult result)
{
var peptideMap = new Dictionary <String, IIdentifiedPeptide>();
foreach (IIdentifiedProteinGroup group in result)
{
foreach (IIdentifiedProtein protein in group)
{
for (int i = 0; i < protein.Peptides.Count; i++)
{
String pepid = protein.Peptides[i].Spectrum.Query.QueryId + "_" + protein.Peptides[i].Sequence;
if (peptideMap.ContainsKey(pepid))
{
IIdentifiedPeptide old = peptideMap[pepid];
old.AddProtein(protein.Name);
protein.Peptides[i] = old;
}
else
{
peptideMap[pepid] = protein.Peptides[i];
}
}
}
}
}
public void RemovePeptideAt(int index)
{
IIdentifiedPeptide peptide = this.peptides[index];
this.peptides.RemoveAt(index);
peptide.Spectrum = null;
}
public void AddPeptide(IIdentifiedPeptide peptide)
{
if (!this.peptides.Contains(peptide))
{
this.peptides.Add(peptide);
peptide.SpectrumBase = this;
}
}
public void RemovePeptide(IIdentifiedPeptide peptide)
{
int index = this.peptides.IndexOf(peptide);
if (index >= 0)
{
RemovePeptideAt(index);
}
}
protected override string DoGetClassification(IIdentifiedPeptide obj)
{
string experimentalName = base.DoGetClassification(obj);
if (!identityMap.ContainsKey(experimentalName))
{
throw new ArgumentException(experimentalName + " does not match to any classified name defined in map, so classification failed!");
}
return(identityMap[experimentalName]);
}
public int CompareTo(IIdentifiedPeptide other)
{
if (this.spectrum == null)
{
return(-1);
}
if (other == null || other.Spectrum == null)
{
return(1);
}
return(this.spectrum.CompareTo(other.Spectrum));
}
private string GetModificationCount(IIdentifiedPeptide pep, string p)
{
var result = 0;
var matchedSeq = PeptideUtils.GetMatchedSequence(pep.Sequence);
for (int i = 0; i < matchedSeq.Length - 1; i++)
{
if (p.Contains(matchedSeq[i]) && !Char.IsLetter(matchedSeq[i + 1]))
{
result++;
i++;
}
}
return(result.ToString());
}
private static void WritePeptide(StreamWriter sw, IIdentifiedSpectrum sph, IIdentifiedPeptide pep)
{
sw.WriteLine(
" <search_hit hit_rank=\"1\" peptide=\"{0}\" peptide_prev_aa=\"{1}\" peptide_next_aa=\"{2}\" protein=\"{3}\" num_tot_proteins=\"{4}\" num_matched_ions=\"{5}\" tot_num_ions=\"{6}\" calc_neutral_pep_mass=\"{7:0.0000}\" massdiff=\"{8}{9:0.0000}\" num_tol_term=\"{10}\" num_missed_cleavages=\"{11}\" is_rejected=\"0\">",
pep.PureSequence,
pep.Sequence[1] == '.' ? pep.Sequence[0] : ' ',
pep.Sequence[pep.Sequence.Length - 2] == '.' ? pep.Sequence[pep.Sequence.Length - 1] : ' ',
pep.Proteins[0],
pep.Proteins.Count,
sph.MatchedIonCount,
sph.TheoreticalIonCount,
sph.TheoreticalMH - Atom.H.AverageMass,
sph.TheoreticalMinusExperimentalMass <= 0 ? "+" : "",
-sph.TheoreticalMinusExperimentalMass,
2,
sph.NumMissedCleavages);
}
protected override bool SkipCurrentEntry(IdentifiedSpectrum entry, IdentifiedSpectrum currEntry)
{
string sNextPureSequence = currEntry.Peptides[0].PureSequence;
for (int i = 0; i < entry.Peptides.Count; i++)
{
IIdentifiedPeptide sp = entry.Peptides[i];
if (sNextPureSequence.Equals(sp.PureSequence))
{
if (currEntry.DeltaScore <= this.maxDeltaScore)
{
entry.DiffModificationSiteCandidates.Add(new FollowCandidate(currEntry.GetSequences(" ! "), currEntry.Score,
currEntry.DeltaScore));
}
return(true);
}
}
return(false);
}
public override void SetProperty(T t, string value)
{
string[] peptides = reg.Split(value);
if (t.Peptides.Count != peptides.Length)
{
t.ClearPeptides();
foreach (string peptide in peptides)
{
IIdentifiedPeptide mp = t.NewPeptide();
mp.Sequence = peptide;
}
}
else
{
for (int i = 0; i < peptides.Length; i++)
{
t.Peptides[i].Sequence = peptides[i];
}
}
}
private static string GetSequenceCharge(IIdentifiedPeptide n)
{
return(n.PureSequence.Replace("I", "L").Replace("Q", "K") + "_" + n.Spectrum.Charge.ToString());
}
private void WriteModificationAndScore(StreamWriter sw, Dictionary <char, double> modChars, IIdentifiedSpectrum sph, IIdentifiedPeptide pep)
{
var matchSeq = PeptideUtils.GetMatchedSequence(pep.Sequence);
if (matchSeq.Any(m => modChars.ContainsKey(m))) // modification
{
if (modChars.ContainsKey(matchSeq[0])) //Nterminal
{
sw.Write(" <modification_info mod_nterm_mass=\"{0:0.######}\"", modChars[matchSeq[0]]);
}
else
{
sw.Write(" <modification_info ");
}
sw.WriteLine(" modified_peptide=\"{0}\">", matchSeq);
int pos = 1;
double mass;
for (int i = 0; i < matchSeq.Length; i++)
{
if (modChars.TryGetValue(matchSeq[i], out mass))
{
if (i == 0)
{
continue;
}
sw.WriteLine(" <mod_aminoacid_mass position=\"{0}\" mass=\"{1:0.######}\"/>", pos, mass);
}
else
{
pos++;
}
}
sw.WriteLine(" </modification_info>");
}
WriteScore(sw, sph);
}
public void Quantify(string rawFileName, List <IIdentifiedSpectrum> spectra, string detailDir)
{
if (!Directory.Exists(detailDir))
{
Directory.CreateDirectory(detailDir);
}
var experimental = RawFileFactory.GetExperimental(rawFileName);
Dictionary <string, DifferentRetentionTimeEnvelopes> spectrumKeyMap = new Dictionary <string, DifferentRetentionTimeEnvelopes>();
Dictionary <SilacEnvelopes, List <IIdentifiedSpectrum> > envelopeSpectrumGroup = new Dictionary <SilacEnvelopes, List <IIdentifiedSpectrum> >();
double precursorPPM = GetPrecursorPPM(spectra);
try
{
_rawReader.Open(rawFileName);
int firstScanNumber = _rawReader.GetFirstSpectrumNumber();
int lastScanNumber = _rawReader.GetLastSpectrumNumber();
Progress.SetRange(1, spectra.Count);
int pepCount = 0;
for (int s = 0; s < spectra.Count; s++)
{
Console.WriteLine(s);
IIdentifiedSpectrum spectrum = spectra[s];
SilacQuantificationSummaryItem.ClearAnnotation(spectrum);
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
int startScan = spectrum.Query.FileScan.FirstScan;
if (startScan > lastScanNumber)
{
spectrum.GetOrCreateQuantificationItem().RatioStr = "OUT_OF_RANGE";
continue;
}
Progress.SetPosition(pepCount++);
IIdentifiedPeptide sp = spectrum.Peptide;
string seq = GetMatchSequence(spectrum);
IPeptideInfo peptideInfo = new IdentifiedPeptideInfo(seq, spectrum.TheoreticalMH, spectrum.Query.Charge);
SilacCompoundInfo sci = GetSilacCompoundInfo(peptideInfo);
//如果轻重离子理论质荷比一样，忽略
if (!sci.IsSilacData())
{
spectrum.GetOrCreateQuantificationItem().RatioStr = "NOT_SILAC";
continue;
}
//如果轻重离子理论质荷比与观测值不一致，忽略
if (!sci.IsMzEquals(spectrum.ObservedMz, MAX_DELTA_MZ))
{
ValidateModifications(seq);
spectrum.GetOrCreateQuantificationItem().RatioStr = "WRONG_IDENTIFICATION";
continue;
}
//如果没有找到相应的FullScan，忽略
int identifiedFullScan = _rawReader.FindPreviousFullScan(startScan, firstScanNumber);
if (-1 == identifiedFullScan)
{
spectrum.GetOrCreateQuantificationItem().RatioStr = "NO_PROFILE";
continue;
}
DifferentRetentionTimeEnvelopes pkls = FindEnvelopes(spectrumKeyMap, spectrum, sci);
SilacEnvelopes envelope = pkls.FindSilacEnvelope(identifiedFullScan);
//如果该scan被包含在已经被定量的结果中，忽略
if (envelope != null)
{
envelope.SetScanIdentified(identifiedFullScan, spectrum.IsExtendedIdentification());
envelopeSpectrumGroup[envelope].Add(spectrum);
continue;
}
//从原始文件中找出该spectrum的定量信息
int maxIndex = Math.Min(option.ProfileLength - 1, pkls.LightProfile.FindMaxIndex());
double mzTolerance = PrecursorUtils.ppm2mz(sci.Light.Mz, option.PPMTolerance);
//如果FullScan没有相应的离子，忽略。（鉴定错误或者扩展定量时候，会出现找不到pair的现象）
SilacPeakListPair splp = GetLightHeavyPeakList(_rawReader, sci, maxIndex, mzTolerance, identifiedFullScan);
if (null == splp)
{
spectrum.GetOrCreateQuantificationItem().RatioStr = "NO_PROFILE";
continue;
}
splp.IsIdentified = true;
splp.IsExtendedIdentification = spectrum.IsExtendedIdentification();
SilacEnvelopes envelopes = new SilacEnvelopes();
envelopes.Add(splp);
//向前查找定量信息
int fullScan = identifiedFullScan;
int scanNumber = 0;
while ((fullScan = _rawReader.FindPreviousFullScan(fullScan - 1, firstScanNumber)) != -1)
{
if (_rawReader.IsBadDataScan(fullScan))
{
continue;
}
scanNumber++;
var item = GetLightHeavyPeakList(_rawReader, sci, maxIndex, mzTolerance, fullScan, scanNumber <= MinScanNumber);
if (null == item)
{
break;
}
envelopes.Add(item);
}
envelopes.Reverse();
//向后查找定量信息
fullScan = identifiedFullScan;
scanNumber = 0;
while ((fullScan = _rawReader.FindNextFullScan(fullScan + 1, lastScanNumber)) != -1)
{
if (_rawReader.IsBadDataScan(fullScan))
{
continue;
}
scanNumber++;
var item = GetLightHeavyPeakList(_rawReader, sci, maxIndex, mzTolerance, fullScan, scanNumber <= MinScanNumber);
if (null == item)
{
break;
}
envelopes.Add(item);
}
//对每个scan计算轻重的离子丰度
envelopes.ForEach(m => m.CalculateIntensity(pkls.LightProfile, pkls.HeavyProfile));
pkls.Add(envelopes);
envelopeSpectrumGroup.Add(envelopes, new List <IIdentifiedSpectrum>());
envelopeSpectrumGroup[envelopes].Add(spectrum);
}
}
finally
{
_rawReader.Close();
}
foreach (string key in spectrumKeyMap.Keys)
{
DifferentRetentionTimeEnvelopes pkls = spectrumKeyMap[key];
foreach (SilacEnvelopes envelopes in pkls)
{
if (0 == envelopes.Count)
{
continue;
}
List <IIdentifiedSpectrum> mps = envelopeSpectrumGroup[envelopes];
double mzTolerance = PrecursorUtils.ppm2mz(mps[0].Query.ObservedMz, option.PPMTolerance);
string scanStr = GetScanRange(envelopes);
string resultFilename = detailDir + "\\" + mps[0].Query.FileScan.Experimental + "." + PeptideUtils.GetPureSequence(mps[0].Sequence) + "." + mps[0].Query.Charge + scanStr + ".silac";
IPeptideInfo peptideInfo = new IdentifiedPeptideInfo(mps[0].GetMatchSequence(), mps[0].TheoreticalMH, mps[0].Query.Charge);
SilacCompoundInfo sci = GetSilacCompoundInfo(peptideInfo);
SilacQuantificationSummaryItem item = new SilacQuantificationSummaryItem(sci.Light.IsSample);
item.RawFilename = rawFileName;
item.SoftwareVersion = this.SoftwareVersion;
item.PeptideSequence = mps[0].Sequence;
item.Charge = mps[0].Charge;
item.LightAtomComposition = sci.Light.Composition.ToString();
item.HeavyAtomComposition = sci.Heavy.Composition.ToString();
item.LightProfile = pkls.LightProfile;
item.HeavyProfile = pkls.HeavyProfile;
item.ObservedEnvelopes = envelopes;
item.ValidateScans(sci, precursorPPM);
item.Smoothing();
item.CalculateRatio();
new SilacQuantificationSummaryItemXmlFormat().WriteToFile(resultFilename, item);
int maxScoreItemIndex = FindMaxScoreItemIndex(mps);
for (int i = 0; i < mps.Count; i++)
{
if (maxScoreItemIndex == i)
{
item.AssignToAnnotation(mps[i], resultFilename);
}
else
{
item.AssignDuplicationToAnnotation(mps[i], resultFilename);
}
}
}
}
foreach (IIdentifiedSpectrum mph in spectra)
{
mph.InitializeRatioEnabled();
}
}
public override IEnumerable <string> Process(string optionFile)
{
this.options = O18QuantificationFileProcessorOptions.Load(optionFile);
var calc = options.GetProteinRatioCalculator();
var detailDirectory = options.GetDetailDirectory();
if (!Directory.Exists(detailDirectory))
{
Directory.CreateDirectory(detailDirectory);
}
var format = new MascotResultTextFormat();
IIdentifiedResult mr = format.ReadFromFile(options.ProteinFile);
CheckRawFilename(mr, optionFile);
Dictionary <string, List <IIdentifiedSpectrum> > filePepMap = GetFilePeptideMap(mr);
Dictionary <IIdentifiedPeptide, O18QuantificationSummaryItem> pepResultMap = new Dictionary <IIdentifiedPeptide, O18QuantificationSummaryItem>();
foreach (string filename in filePepMap.Keys)
{
Progress.SetMessage("Processing " + filename);
string rawFilename = filename;
if (new FileInfo(filename).Name.Equals("Cmpd.raw"))
{
rawFilename = FindRawFileName(options.ProteinFile);
}
string experimental = FileUtils.ChangeExtension(new FileInfo(rawFilename).Name, "");
using (CacheRawFile rawFile = new CacheRawFile(rawFilename))
{
int firstScanNumber = rawFile.GetFirstSpectrumNumber();
int lastScanNumber = rawFile.GetLastSpectrumNumber();
List <IIdentifiedSpectrum> peps = filePepMap[filename];
Dictionary <string, DifferentRetentionTimeEnvelopes> peptideChargeMap = new Dictionary <string, DifferentRetentionTimeEnvelopes>();
Dictionary <O18QuanEnvelopes, List <IIdentifiedSpectrum> > pklMpMap = new Dictionary <O18QuanEnvelopes, List <IIdentifiedSpectrum> >();
Progress.SetRange(0, peps.Count);
foreach (IIdentifiedSpectrum mphit in peps)
{
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
Progress.Increment(1);
IIdentifiedPeptide mp = mphit.Peptide;
if (mp.Sequence.EndsWith("-"))
{
//it cannot be O18 labelled, just skip it
continue;
}
int startScan = mphit.Query.FileScan.FirstScan;
double theoreticalMz = GetTheoretialO16Mz(gapO18O16, mphit);
int theoreticalMass = (int)(theoreticalMz * mphit.Query.Charge + 0.5);
string sequenceCharge = PeptideUtils.GetPureSequence(mphit.Sequence) + "." + mphit.Query.Charge + "." + theoreticalMass;
if (!peptideChargeMap.ContainsKey(sequenceCharge))
{
peptideChargeMap.Add(sequenceCharge, new DifferentRetentionTimeEnvelopes());
}
bool bFound = false;
DifferentRetentionTimeEnvelopes pkls = peptideChargeMap[sequenceCharge];
foreach (var pklList in pkls)
{
if (pklList.Count == 0)
{
continue;
}
if (pklList[0].Scan > startScan)
{
continue;
}
if (pklList[pklList.Count - 1].Scan < startScan)
{
continue;
}
pklMpMap[pklList].Add(mphit);
bFound = true;
bool findIdentified = false;
for (int i = 1; i < pklList.Count; i++)
{
if (pklList[i].ScanTimes[0].Scan > startScan)
{
pklList[i - 1].IsIdentified = true;
findIdentified = true;
break;
}
}
if (!findIdentified)
{
pklList[pklList.Count - 1].IsIdentified = true;
}
}
if (bFound)
{
continue;
}
double mzTolerance = PrecursorUtils.ppm2mz(theoreticalMz, options.PPMTolerance);
O18QuanEnvelopes envelopes = new O18QuanEnvelopes();
bool bFirst = true;
int count = 0;
//backward
for (int scan = startScan; scan >= firstScanNumber; scan--)
{
if (1 == rawFile.GetMsLevel(scan))
{
O18QuanEnvelope envelope = GetCorrespondingEnvelope(rawFile, theoreticalMz, mphit.Query.Charge, mzTolerance, scan);
//At most one invalid scan inside both pre or post identification scan range.
if (!IsValidEnvelope(envelope, mphit.Charge))
{
if (count > 0)
{
envelopes.RemoveAt(0);
break;
}
else
{
count++;
}
}
else
{
count = 0;
}
if (bFirst)
{
envelope.IsIdentified = true;
bFirst = false;
}
envelopes.Insert(0, envelope);
}
}
if (envelopes.Count == 0)
{
//If the identified scan has no quantification information ,ignore it.
continue;
}
count = 0;
//forward
for (int scan = startScan + 1; scan <= lastScanNumber; scan++)
{
if (1 == rawFile.GetMsLevel(scan))
{
var envelope = GetCorrespondingEnvelope(rawFile, theoreticalMz, mphit.Query.Charge, mzTolerance, scan);
//At most one invalid scan inside both pre or post identification scan range.
if (!IsValidEnvelope(envelope, mphit.Charge))
{
if (count > 0)
{
envelopes.RemoveAt(envelopes.Count - 1);
break;
}
else
{
count = 1;
}
}
else
{
count = 0;
}
envelopes.Add(envelope);
}
}
if (envelopes.Count == 0)
{
continue;
}
string scanCurr = envelopes.GetScanRange();
//check scan list again
bFound = false;
foreach (var pklList in pkls)
{
if (pklList.Count == 0)
{
continue;
}
string scanOld = pklList.GetScanRange();
if (scanOld.Equals(scanCurr))
{
pklMpMap[pklList].Add(mphit);
bFound = true;
break;
}
}
if (bFound)
{
continue;
}
pkls.Add(envelopes);
pklMpMap.Add(envelopes, new List <IIdentifiedSpectrum>());
pklMpMap[envelopes].Add(mphit);
}
var detailFilePrefix = options.GetDetailDirectory() + "\\" + new FileInfo(options.ProteinFile).Name;
foreach (string sequenceCharge in peptideChargeMap.Keys)
{
DifferentRetentionTimeEnvelopes pkls = peptideChargeMap[sequenceCharge];
foreach (var envelopes in pkls)
{
if (0 == envelopes.Count)
{
continue;
}
List <IIdentifiedSpectrum> mps = pklMpMap[envelopes];
double mzTolerance = PrecursorUtils.ppm2mz(mps[0].Query.ObservedMz, options.PPMTolerance);
O18QuantificationPeptideProcessor processor = new O18QuantificationPeptideProcessor(fileFormat,
options.IsPostDigestionLabelling,
rawFilename,
PeptideUtils.GetPureSequence(mps[0].Sequence),
options.PurityOfO18Water,
envelopes, mzTolerance,
"",
options.GetScanPercentageStart() / 100,
options.GetScanPercentageEnd() / 100);
processor.TheoreticalMz = GetTheoretialO16Mz(gapO18O16, mps[0]);
processor.Charge = mps[0].Charge;
processor.SoftwareVersion = options.SoftwareVersion;
var resultFilename = MyConvert.Format("{0}.{1}.{2}.{3}.{4}.O18",
detailFilePrefix,
experimental,
PeptideUtils.GetPureSequence(mps[0].Sequence),
mps[0].Charge,
envelopes.GetScanRange());
processor.Process(resultFilename);
O18QuantificationSummaryItem item = fileFormat.ReadFromFile(resultFilename);
int maxScoreItemIndex = FindMaxScoreItemIndex(mps);
var relativeFile = Path.Combine(Path.GetFileName(options.GetDetailDirectory()), Path.GetFileName(resultFilename));
for (int i = 0; i < mps.Count; i++)
{
if (maxScoreItemIndex == i)
{
item.AssignToAnnotation(mps[i], relativeFile);
}
else
{
item.AssignDuplicationToAnnotation(mps[i], relativeFile);
}
}
}
}
}
}
List <IIdentifiedSpectrum> peptides = mr.GetSpectra();
foreach (IIdentifiedSpectrum mphit in peptides)
{
if (!mphit.Annotations.ContainsKey(O18QuantificationConstants.O18_RATIO_SCANCOUNT))
{
mphit.Annotations[O18QuantificationConstants.O18_RATIO_SCANCOUNT] = "-";
}
mphit.SetEnabled(calc.HasPeptideRatio(mphit));
}
calc.Calculate(mr, m => true);
string resultFile = FileUtils.ChangeExtension(optionFile, ".O18summary");
format.InitializeByResult(mr);
format.ProteinFormat = format.ProteinFormat.GetLineFormat(O18QuantificationConstants.O18_EXPORT_PROTEIN_HEADER);
format.WriteToFile(resultFile, mr);
Progress.SetMessage("Finished, result was saved to " + resultFile);
return(new[] { resultFile });
}
