private void mnuOpen_Click(object sender, EventArgs e)
{
var dlg = new ITraqOpenFileDialog();
dlg.LoadOption();
if (dlg.ShowDialog(this) == System.Windows.Forms.DialogResult.OK)
{
proteinFile = dlg.ProteinFile;
itrapFile = dlg.ITraqFile;
var paramFile = Path.ChangeExtension(dlg.ProteinFile, "param");
if (!File.Exists(paramFile))
{
MessageBox.Show(this, "Error find parameter file " + paramFile);
return;
}
option = new ITraqProteinStatisticOption();
option.LoadFromFile(paramFile);
Progress.Begin();
var task = Task.Factory.StartNew(() => LoadData(), TaskCreationOptions.LongRunning);
task.ContinueWith((m) => this.Invoke(new Action(UpdateData)), TaskContinuationOptions.OnlyOnRanToCompletion);
}
}
protected List <PeakList <Peak> > MergePeakList(List <PeakList <Peak> > pklList)
{
int index = 0;
Progress.SetRange(0, pklList.Count);
Progress.Begin();
try
{
while (index < pklList.Count)
{
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
Progress.SetPosition(index);
PeakList <Peak> currentPkl = pklList[index];
double maxGap = PrecursorUtils.ppm2mz(currentPkl.PrecursorMZ, this.ppmPrecursorTolerance);
int next = index + 1;
while (next < pklList.Count)
{
PeakList <Peak> nextPkl = pklList[next];
double retentionTimeGap = nextPkl.ScanTimes[0].RetentionTime -
currentPkl.ScanTimes[0].RetentionTime;
if (retentionTimeGap > this.retentionTimeTolerance)
{
break;
}
if (nextPkl.PrecursorCharge != currentPkl.PrecursorCharge)
{
next++;
continue;
}
double precursorMzGap = Math.Abs(nextPkl.PrecursorMZ - currentPkl.PrecursorMZ);
if (precursorMzGap < maxGap)
{
currentPkl.MergeByMZFirst(nextPkl, this.ppmPeakTolerance);
pklList.RemoveAt(next);
continue;
}
next++;
}
index++;
}
Progress.SetPosition(pklList.Count);
}
finally
{
Progress.End();
}
return(pklList);
}
public override IEnumerable <string> Process()
{
var expRawfileMap = options.RawFiles.ToDictionary(m => Path.GetFileNameWithoutExtension(m));
Progress.SetMessage("Reading library file ...");
var liblist = new MS2ItemXmlFormat().ReadFromFile(options.LibraryFile);
PreprocessingMS2ItemList(liblist);
var lib = liblist.GroupBy(m => m.Charge).ToDictionary(m => m.Key, m => m.ToList());
Progress.SetMessage("Building library sequence amino acid composition ...");
lib.ForEach(m => m.Value.ForEach(l => l.AminoacidCompsition = (from a in l.Peptide
where options.SubstitutionDeltaMassMap.ContainsKey(a)
select a).Distinct().OrderBy(k => k).ToArray()));
var expScanMap = (from p in liblist
from sq in p.FileScans
select sq).ToList().GroupBy(m => m.Experimental).ToDictionary(m => m.Key, m => new HashSet <int>(from l in m select l.FirstScan));
if (File.Exists(options.PeptidesFile))
{
Progress.SetMessage("Reading peptides file used for excluding scan ...");
var peptides = new MascotPeptideTextFormat().ReadFromFile(options.PeptidesFile);
foreach (var pep in peptides)
{
HashSet <int> scans;
if (!expScanMap.TryGetValue(pep.Query.FileScan.Experimental, out scans))
{
scans = new HashSet <int>();
expScanMap[pep.Query.FileScan.Experimental] = scans;
}
scans.Add(pep.Query.FileScan.FirstScan);
}
}
Progress.SetMessage("Reading MS2/MS3 data ...");
var result = GetCandidateMs2ItemList(expRawfileMap, expScanMap);
PreprocessingMS2ItemList(result);
//new MS2ItemXmlFormat().WriteToFile(options.OutputFile + ".xml", result);
Progress.SetMessage("Finding SAP ...");
List <SapPredicted> predicted = new List <SapPredicted>();
var minDeltaMass = options.SubstitutionDeltaMassMap.Values.Min(l => l.Min(k => k.DeltaMass));
var maxDeltaMass = options.SubstitutionDeltaMassMap.Values.Max(l => l.Max(k => k.DeltaMass));
Progress.SetRange(0, result.Count);
Progress.Begin();
FindCandidates(lib, result, predicted, minDeltaMass, maxDeltaMass);
var groups = predicted.ToGroupDictionary(m => m.Ms2.GetFileScans());
predicted.Clear();
foreach (var g in groups.Values)
{
var gg = g.ToGroupDictionary(m => m.LibMs2).Values.ToList();
gg.Sort((m1, m2) =>
{
return(CompareSapPrecitedList(m1, m2));
});
var expect = gg[0].FirstOrDefault(m => m.IsExpect);
if (expect != null)
{
predicted.Add(expect);
}
else
{
predicted.AddRange(gg[0]);
for (int i = 1; i < gg.Count; i++)
{
if (CompareSapPrecitedList(gg[0], gg[i]) == 0)
{
predicted.AddRange(gg[i]);
}
else
{
break;
}
}
}
}
if (File.Exists(options.MatchedFile))
{
new SapPredictedValidationWriter(options.MatchedFile).WriteToFile(options.OutputFile, predicted);
}
else
{
new SapPredictedWriter().WriteToFile(options.OutputTableFile, predicted);
Progress.SetMessage("Generating SAP sequence ...");
List <Sequence> predictedSeq = new List <Sequence>();
foreach (var predict in predicted)
{
var seq = PeptideUtils.GetPureSequence(predict.LibMs2.Peptide);
if (predict.Target.TargetType == VariantType.SingleAminoacidPolymorphism)
{
for (int i = 0; i < seq.Length; i++)
{
if (seq[i] == predict.Target.Source[0])
{
foreach (var t in predict.Target.Target)
{
string targetSeq;
if (i == 0)
{
targetSeq = t + seq.Substring(1);
}
else
{
targetSeq = seq.Substring(0, i) + t + seq.Substring(i + 1);
}
var reference = string.Format("sp|SAP_{0}_{1}|{2}_{3}_{4}_{5}", targetSeq, predict.Target.TargetType, seq, predict.Target.Source, i + 1, t);
predictedSeq.Add(new Sequence(reference, targetSeq));
}
}
}
}
else
{
foreach (var tseq in predict.Target.Target)
{
string reference;
if (predict.Target.TargetType == VariantType.NTerminalLoss)
{
reference = string.Format("sp|SAP_{0}_{1}|{2}_loss_{3}", tseq, predict.Target.TargetType, seq, seq.Substring(0, seq.Length - tseq.Length));
}
else if (predict.Target.TargetType == VariantType.CTerminalLoss)
{
reference = string.Format("sp|SAP_{0}_{1}|{2}_loss_{3}", tseq, predict.Target.TargetType, seq, seq.Substring(tseq.Length));
}
else if (predict.Target.TargetType == VariantType.NTerminalExtension)
{
reference = string.Format("sp|SAP_{0}_{1}|{2}_ext_{3}", tseq, predict.Target.TargetType, seq, tseq.Substring(0, tseq.Length - seq.Length));
}
else if (predict.Target.TargetType == VariantType.CTerminalExtension)
{
reference = string.Format("sp|SAP_{0}_{1}|{2}_ext_{3}", tseq, predict.Target.TargetType, seq, tseq.Substring(seq.Length));
}
else
{
throw new Exception("I don't know how to deal with " + predict.Target.TargetType.ToString());
}
predictedSeq.Add(new Sequence(reference, tseq));
}
}
}
predictedSeq = (from g in predictedSeq.GroupBy(m => m.SeqString)
select g.First()).ToList();
Progress.SetMessage("Reading database {0} ...", options.DatabaseFastaFile);
var databases = SequenceUtils.Read(options.DatabaseFastaFile);
Progress.SetMessage("Removing variant sequences which are already existed in database ...");
for (int i = predictedSeq.Count - 1; i >= 0; i--)
{
foreach (var db in databases)
{
if (db.SeqString.Contains(predictedSeq[i].SeqString))
{
predictedSeq.RemoveAt(i);
break;
}
}
}
databases.AddRange(predictedSeq);
Progress.SetMessage("Writing SAP sequence and original database to {0} ...", options.OutputFile);
SequenceUtils.Write(new FastaFormat(), options.OutputFile, databases);
}
Progress.End();
return(new string[] { options.OutputFile, options.OutputTableFile });
}
public void Begin()
{
Progress.Begin();
}
protected List <PeakList <Peak> > ReadTandemMassFromRaw(FileInfo rawFilename, List <int> ignoreScans)
{
string experimental = FileUtils.ChangeExtension(rawFilename.Name, "");
var result = new List <PeakList <Peak> >();
bool bReadAgain = false;
rawReader.Open(rawFilename.FullName);
try
{
int firstSpectrumNumber = rawReader.GetFirstSpectrumNumber();
int lastSpectrumNumber = rawReader.GetLastSpectrumNumber();
Progress.SetRange(firstSpectrumNumber, lastSpectrumNumber);
Progress.Begin();
try
{
for (int scan = firstSpectrumNumber; scan <= lastSpectrumNumber; scan++)
{
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
if (ignoreScans.Contains(scan))
{
continue;
}
Progress.SetPosition(scan);
int msLevel = rawReader.GetMsLevel(scan);
if (msLevel > 1)
{
PeakList <Peak> pkl;
try
{
pkl = rawReader.GetPeakList(scan);
}
catch (RawReadException ex)
{
ignoreScans.Add(ex.Scan);
File.WriteAllLines(GetIgnoreScanFile(rawFilename), (from i in ignoreScans
let s = i.ToString()
select s).ToArray());
bReadAgain = true;
break;
}
pkl.Precursor = rawReader.GetPrecursorPeakWithMasterScan(scan);
pkl.MsLevel = msLevel;
pkl.Experimental = experimental;
pkl.ScanTimes.Add(new ScanTime(scan, rawReader.ScanToRetentionTime(scan)));
pkl.ScanMode = rawReader.GetScanMode(scan);
if (pkl.PrecursorCharge == 0)
{
pkl.PrecursorCharge = PrecursorUtils.GuessPrecursorCharge(pkl, pkl.PrecursorMZ);
}
PeakList <Peak> pklProcessed = this.pklProcessor.Process(pkl);
if (null != pklProcessed && pklProcessed.Count > 0)
{
result.Add(pklProcessed);
}
}
}
}
finally
{
Progress.End();
}
}
finally
{
rawReader.Close();
}
if (bReadAgain)
{
return(ReadTandemMassFromRaw(rawFilename, ignoreScans));
}
else
{
return(result);
}
}