/// <summary>
/// 读取fasta文件，进行数据处理。
/// </summary>
/// <param name="fileName"></param>
/// <returns></returns>
public override IEnumerable <string> Process(string fastaFile)
{
HashSet <string> result = new HashSet <string>();
var ff = new FastaFormat();
using (StreamReader sr = new StreamReader(fastaFile))
{
Progress.SetRange(0, sr.BaseStream.Length);
var aas = new Aminoacids();
Predicate <string> aaFilter = m =>
{
foreach (var aa in ignoreAminoacids)
{
if (m.Contains(aa))
{
return(false);
}
}
return(true);
};
Predicate <string> lengthFilter = m => m.Length >= minLength;
Predicate <string> massFilter = m =>
{
var mass = aas.MonoPeptideMass(m);
return(mass >= minMass && mass <= maxMass);
};
Predicate <string> filter = m => aaFilter(m) && lengthFilter(m) && massFilter(m);
List <Digest> digs = new List <Digest>();
foreach (var protease in proteases)
{
var dig = new Digest();
dig.DigestProtease = protease;
dig.MaxMissedCleavages = maxMissCleavage;
digs.Add(dig);
}
Sequence seq;
Progress.SetMessage("Digesting sequences ...");
while ((seq = ff.ReadSequence(sr)) != null)
{
Progress.SetPosition(sr.GetCharpos());
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
HashSet <string> curseqs = new HashSet <string>();
curseqs.Add(seq.SeqString);
foreach (var dig in digs)
{
var last = curseqs;
curseqs = new HashSet <string>();
foreach (var curseq in last)
{
var pro = new Sequence(curseq, curseq);
dig.ProteinSequence = pro;
dig.AddDigestFeatures();
var infos = pro.GetDigestPeptideInfo();
infos.ForEach(m =>
{
if (filter(m.PeptideSeq))
{
curseqs.Add(m.PeptideSeq);
}
});
}
}
result.UnionWith(curseqs);
}
}
Progress.SetMessage("Sorting sequences ...");
var peps = new List <string>(result);
peps.Sort((m1, m2) =>
{
var res = m1.Length.CompareTo(m2.Length);
if (res == 0)
{
res = m1.CompareTo(m2);
}
return(res);
});
var resultFile = fastaFile + ".pep";
using (StreamWriter sw = new StreamWriter(resultFile))
{
peps.ForEach(m => sw.WriteLine(m));
}
return(new[] { resultFile });
}
private void iIdentifiedSpectrumDataGridView_SelectionChanged(object sender, EventArgs e)
{
CurrencyManager cm = (CurrencyManager)this.BindingContext[peptides];
var spectrum = cm.Current as IIdentifiedSpectrum;
if (spectrum.Annotations.ContainsKey("PepMutation"))
{
zgcPeaks.MasterPane.Title.Text = string.Format("MH={0:0.00000},PPM={1:0.0},SEQ={2},MUT={3},COUNT={4}",
spectrum.TheoreticalMH,
PrecursorUtils.mz2ppm(spectrum.TheoreticalMH, spectrum.TheoreticalMinusExperimentalMass),
spectrum.Sequence,
spectrum.Annotations["PepMutation"],
(from a in spectrum.Annotations
where a.Key.EndsWith("_PepCount")
select Convert.ToInt32(a.Value)).Sum());
}
else
{
zgcPeaks.MasterPane.Title.Text = string.Format("MH={0:0.00000},PPM={1:0.0},SEQ={2}",
spectrum.TheoreticalMH,
PrecursorUtils.mz2ppm(spectrum.TheoreticalMH, spectrum.TheoreticalMinusExperimentalMass),
spectrum.Sequence);
}
var mass = aas.MonoPeptideMass(PeptideUtils.GetMatchedSequence(spectrum.Sequence));
if (Math.Abs(mass - spectrum.TheoreticalMass) > 0.1)
{
MessageBox.Show(string.Format("Error : {0} - {1}", mass, spectrum.TheoreticalMass));
return;
}
var name = string.Format("{0}_{1}", spectrum.Query.FileScan.Experimental, spectrum.Query.FileScan.FirstScan);
var peakPane = zgcPeaks.MasterPane.PaneList[0];
peakPane.ClearData();
var ppmPane = zgcPeaks.MasterPane.PaneList[1];
ppmPane.ClearData();
if (!mgfMap.ContainsKey(name))
{
MessageBox.Show("Cannot find peak list {0}", name);
return;
}
var mgf = mgfMap[name];
var maxIntensity = mgf.Max(m => m.Intensity);
var maxMz = Math.Min(2000, mgf.Max(m => m.Mz));
var mzTolerance = 0.02;
var minIntensity = maxIntensity * 0.05;
AddIonSeries(peakPane, ppmPane, mgf, mzTolerance, minIntensity, yBuilder, spectrum.Sequence, Color.Red);
AddIonSeries(peakPane, ppmPane, mgf, mzTolerance, minIntensity, bBuilder, spectrum.Sequence, Color.Blue);
if (spectrum.Charge > 2)
{
AddIonSeries(peakPane, ppmPane, mgf, mzTolerance, minIntensity, y2Builder, spectrum.Sequence, Color.Brown);
AddIonSeries(peakPane, ppmPane, mgf, mzTolerance, minIntensity, b2Builder, spectrum.Sequence, Color.GreenYellow);
}
AddUnmatchedIons(peakPane, mgf);
foreach (var pane in zgcPeaks.MasterPane.PaneList)
{
pane.XAxis.Scale.Min = 0;
pane.XAxis.Scale.Max = maxMz;
}
zgcPeaks.UpdateGraph();
}
private void DrawSpectrum(IIdentifiedSpectrum spectrum)
{
if (spectrum.Annotations.ContainsKey("PepMutation"))
{
zgcPeaks.MasterPane.PaneList[0].Title.Text = string.Format("MH={0:0.00000},PPM={1:0.0},SEQ={2},MUT={3},COUNT={4}",
spectrum.TheoreticalMH,
PrecursorUtils.mz2ppm(spectrum.TheoreticalMH, spectrum.TheoreticalMinusExperimentalMass),
spectrum.Sequence,
spectrum.Annotations["PepMutation"],
(from a in spectrum.Annotations
where a.Key.EndsWith("_PepCount")
select Convert.ToInt32(a.Value)).Sum());
}
else
{
zgcPeaks.MasterPane.PaneList[0].Title.Text = string.Format("MH={0:0.00000},PPM={1:0.0},SEQ={2}",
spectrum.TheoreticalMH,
PrecursorUtils.mz2ppm(spectrum.TheoreticalMH, spectrum.TheoreticalMinusExperimentalMass),
spectrum.Sequence);
}
var mass = aas.MonoPeptideMass(PeptideUtils.GetMatchedSequence(spectrum.Sequence));
if (Math.Abs(mass - spectrum.TheoreticalMass) > 0.1)
{
MessageBox.Show(string.Format("Error : {0} - {1}", mass, spectrum.TheoreticalMass));
return;
}
var name = string.Format("{0}_{1}", spectrum.Query.FileScan.Experimental, spectrum.Query.FileScan.FirstScan);
var peakPane = zgcPeaks.MasterPane.PaneList[0];
peakPane.ClearData();
var ppmPane = zgcPeaks.MasterPane.PaneList[1];
ppmPane.ClearData();
if (!mgfMap.ContainsKey(name))
{
MessageBox.Show("Cannot find peak list {0}", name);
return;
}
var mgf = mgfMap[name];
var maxIntensity = mgf.Max(m => m.Intensity);
var maxMz = Math.Min(2000, mgf.Max(m => m.Mz));
var productTolerance = productIonPPM;
var minIntensity = maxIntensity * 0.05;
AddIonSeries(peakPane, ppmPane, mgf, productTolerance, minIntensity, yBuilder, spectrum.Sequence, Color.Red);
AddIonSeries(peakPane, ppmPane, mgf, productTolerance, minIntensity, bBuilder, spectrum.Sequence, Color.Blue);
if (spectrum.Charge > 2)
{
AddIonSeries(peakPane, ppmPane, mgf, productTolerance, minIntensity, y2Builder, spectrum.Sequence, Color.Brown);
AddIonSeries(peakPane, ppmPane, mgf, productTolerance, minIntensity, b2Builder, spectrum.Sequence, Color.GreenYellow);
}
AddUnmatchedIons(peakPane, mgf);
foreach (var pane in zgcPeaks.MasterPane.PaneList)
{
pane.XAxis.Scale.Min = 0;
pane.XAxis.Scale.Max = maxMz;
}
zgcPeaks.UpdateGraph();
var targetpngfile = GetTargetFile(targetDir, spectrum);
zgcPeaks.GetImage().Save(targetpngfile, ImageFormat.Png);
}