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()
{
var boundaryInput = Path.ChangeExtension(options.OutputFile, ".chros.tsv");
if (!File.Exists(boundaryInput) || options.Overwrite)
{
var format = GetPeptideReader();
var spectra = format.ReadFromFile(options.InputFile);
var peptideMap = spectra.ToGroupDictionary(m => m.Query.FileScan.Experimental.ToLower());
var rawfiles = options.RawFiles.ToDictionary(m => RawFileFactory.GetExperimental(m).ToLower());
var rententionWindow = options.MaximumRetentionTimeWindow;
var missed = peptideMap.Keys.Except(rawfiles.Keys).ToArray();
if (missed.Length > 0)
{
throw new Exception(string.Format("Cannot find raw file of {0} in file list", missed.Merge("/")));
}
var optionThreadCount = options.ThreadCount == 0 ? Environment.ProcessorCount : options.ThreadCount;
var option = new ParallelOptions()
{
MaxDegreeOfParallelism = Math.Min(optionThreadCount, peptideMap.Count),
};
var chroMap = new List <Tuple <string, List <ChromatographProfile> > >();
foreach (var raw in peptideMap)
{
var peptides = raw.Value;
var waitingPeaks = new List <ChromatographProfile>();
foreach (var peptide in peptides)
{
var chro = new ChromatographProfile()
{
Experimental = peptide.Query.FileScan.Experimental,
IdentifiedScan = peptide.Query.FileScan.FirstScan,
IdentifiedRetentionTime = peptide.Query.FileScan.RetentionTime,
ObservedMz = peptide.GetPrecursorMz(),
TheoreticalMz = peptide.GetTheoreticalMz(),
Charge = peptide.Query.Charge,
Sequence = peptide.Peptide.PureSequence,
FileName = GetTargetFile(peptide),
SubFileName = GetTargetSubFile(peptide)
};
chro.InitializeIsotopicIons(options.MzTolerancePPM, options.MinimumIsotopicPercentage);
waitingPeaks.Add(chro);
}
chroMap.Add(new Tuple <string, List <ChromatographProfile> >(raw.Key, waitingPeaks));
}
ConcurrentBag <ChromatographProfile> detected = new ConcurrentBag <ChromatographProfile>();
Parallel.ForEach(chroMap, option, raw =>
{
var rawFileName = raw.Item1;
var waitingPeaks = raw.Item2;
Dictionary <string, List <ChromatographProfile> > resultMap = new Dictionary <string, List <ChromatographProfile> >();
List <FullMS> fullMSList = new List <FullMS>();
Progress.SetMessage("Reading full ms list from " + rawfiles[rawFileName] + "...");
using (var rawReader = new CacheRawFile(RawFileFactory.GetRawFileReader(rawfiles[rawFileName])))
{
var firstScan = rawReader.GetFirstSpectrumNumber();
var lastScan = rawReader.GetLastSpectrumNumber();
for (int scan = firstScan; scan <= lastScan; scan++)
{
var mslevel = rawReader.GetMsLevel(scan);
if (mslevel == 1)
{
fullMSList.Add(new FullMS()
{
Scan = scan,
RetentionTime = rawReader.ScanToRetentionTime(scan),
Peaks = null
});
}
}
foreach (var chro in waitingPeaks)
{
if (chro.IdentifiedScan == 0 && chro.IdentifiedRetentionTime > 0)
{
for (int i = 1; i < fullMSList.Count; i++)
{
if (chro.IdentifiedRetentionTime < fullMSList[i].RetentionTime)
{
break;
}
chro.IdentifiedScan = fullMSList[i].Scan + 1;
}
}
}
var chroGroups = waitingPeaks.GroupBy(chro => chro.GetPeptideId());
foreach (var chroGroup in chroGroups)
{
List <ChromatographProfile> profileChros = new List <ChromatographProfile>();
foreach (var chro in chroGroup.OrderBy(m => m.IdentifiedScan))
{
var masterScanIndex = 0;
for (int i = 1; i < fullMSList.Count; i++)
{
if (chro.IdentifiedScan < fullMSList[i].Scan)
{
break;
}
masterScanIndex = i;
}
var masterScan = fullMSList[masterScanIndex].Scan;
var masterRetentionTime = fullMSList[masterScanIndex].RetentionTime;
bool bExist = false;
foreach (var profileChro in profileChros)
{
foreach (var pkl in profileChro.Profiles)
{
if (pkl.Scan == fullMSList[masterScanIndex].Scan)
{
pkl.Identified = true;
bExist = true;
break;
}
}
if (bExist)
{
break;
}
}
if (bExist)
{
continue;
}
//Progress.SetMessage("Processing {0} : {1:0.#####} : {2} : {3}", chro.Sequence, chro.ObservedMz, chro.IdentifiedScan, Path.GetFileName(chro.FileName));
//allow one missed scan
int naCount = 2;
for (int scanIndex = masterScanIndex; scanIndex >= 0; scanIndex--)
{
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
var curRetentionTime = fullMSList[scanIndex].RetentionTime;
if (masterRetentionTime - curRetentionTime > rententionWindow)
{
break;
}
if (!AddEnvelope(chro, rawReader, fullMSList, scanIndex))
{
naCount--;
if (naCount == 0)
{
break;
}
else
{
continue;
}
}
if (scanIndex == masterScanIndex)
{
chro.Profiles.Last().Identified = true;
}
}
chro.Profiles.Reverse();
naCount = 2;
for (int scanIndex = masterScanIndex + 1; scanIndex < fullMSList.Count; scanIndex++)
{
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
var curRetentionTime = fullMSList[scanIndex].RetentionTime;
if (curRetentionTime - masterRetentionTime > rententionWindow)
{
break;
}
if (!AddEnvelope(chro, rawReader, fullMSList, scanIndex))
{
naCount--;
if (naCount == 0)
{
break;
}
else
{
continue;
}
}
}
profileChros.Add(chro);
}
profileChros.RemoveAll(l => l.Profiles.Count < options.MinimumScanCount);
profileChros.Sort((m1, m2) => m2.Profiles.Count.CompareTo(m1.Profiles.Count));
bool bMain = true;
foreach (var chro in profileChros)
{
var filename = bMain ? chro.FileName : chro.SubFileName;
if (bMain)
{
detected.Add(chro);
}
bMain = false;
new ChromatographProfileTextWriter().WriteToFile(filename, chro);
new ChromatographProfileXmlFormat().WriteToFile(filename + ".xml", chro);
}
}
}
}
);
var chroList = new List <ChromatographProfile>(detected);
chroList.Sort((m1, m2) => m1.FileName.CompareTo(m2.FileName));
if (chroList.Count == 0)
{
throw new Exception("Cannot find chromotograph!");
}
using (var sw = new StreamWriter(boundaryInput))
{
sw.WriteLine("ChroDirectory\tChroFile\tSample\tPeptideId\tTheoreticalMz\tCharge\tIdentifiedScan");
foreach (var chro in chroList)
{
sw.WriteLine("{0}\t{1}\t{2}\t{3}\t{4}\t{5}\t{6}",
Path.GetDirectoryName(chro.FileName).Replace("\\", "/"),
Path.GetFileNameWithoutExtension(chro.FileName),
chro.Experimental,
chro.GetPeptideId(),
chro.TheoreticalMz,
chro.Charge,
chro.IdentifiedScan);
}
}
}
if (!File.Exists(options.OutputFile) || options.Overwrite)
{
Progress.SetMessage("Finding boundaries ...");
var boundaryOptions = new RTemplateProcessorOptions()
{
InputFile = boundaryInput,
OutputFile = options.OutputFile,
RTemplate = BoundaryR,
RExecute = ExternalProgramConfig.GetExternalProgram("R"),
CreateNoWindow = true
};
boundaryOptions.Parameters.Add("outputImage<-" + (options.DrawImage ? "1" : "0"));
boundaryOptions.Parameters.Add("maximumProfileDistance<-" + options.MaximumProfileDistance.ToString());
new RTemplateProcessor(boundaryOptions)
{
Progress = this.Progress
}.Process();
}
//if (options.DrawImage)
//{
// Progress.SetMessage("Drawing images ...");
// var imageOptions = new RTemplateProcessorOptions()
// {
// InputFile = options.OutputFile,
// OutputFile = Path.ChangeExtension(options.OutputFile, ".image"),
// RTemplate = ImageR,
// RExecute = SystemUtils.GetRExecuteLocation(),
// CreateNoWindow = true,
// NoResultFile = true
// };
// new RTemplateProcessor(imageOptions) { Progress = this.Progress }.Process();
//}
return(new string[] { options.OutputFile });
}