public Dictionary <int, List <IIdentifiedSpectrum> > DoParsePeptides(string datFilename, int minRank, double minScore, bool isDecoy)
{
var result = new Dictionary <int, List <IIdentifiedSpectrum> >();
Dictionary <string, string> headers;
int queryCount;
Dictionary <int, MascotQueryItem> queryItems;
Dictionary <string, string> peptideSection;
var prefix = isDecoy ? "decoy_" : "";
using (var sr = new StreamReader(datFilename))
{
InitializeBoundary(sr);
CurrentParameters = ParseSection(sr, "parameters");
var hasDecoy = CurrentParameters.ContainsKey("DECOY") && CurrentParameters["DECOY"].Equals("1");
if (!hasDecoy && isDecoy)
{
return(result);
}
var masses = ParseSection(sr, "masses");
CurrentModifications = ParseModification(masses);
long curPos = sr.GetCharpos();
CurrentProtease = ParseEnzyme(sr);
sr.SetCharpos(curPos);
headers = ParseSection(sr, "header");
queryCount = int.Parse(headers["queries"]);
queryItems = ParseQueryItems(sr, queryCount, prefix);
peptideSection = ParseSection(sr, prefix + "peptides", !isDecoy);
}
string file = CurrentParameters["FILE"];
if (file.StartsWith("File Name: "))
{
file = file.Substring(10).Trim();
}
string defaultExperimental = FileUtils.ChangeExtension(new FileInfo(file).Name, "");
bool isPrecursorMonoisotopic = true;
if (CurrentParameters.ContainsKey("MASS"))
{
isPrecursorMonoisotopic = CurrentParameters["MASS"].Equals("Monoisotopic");
}
using (var sr = new StreamReader(datFilename))
{
//Progress.SetRange(1, queryCount);
for (int queryId = 1; queryId <= queryCount; queryId++)
{
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
//Progress.SetPosition(queryId);
MascotQueryItem queryItem = queryItems[queryId];
var iPeps = new List <IIdentifiedSpectrum>();
result[queryId] = iPeps;
IIdentifiedSpectrum lastHit = null;
int rank = 0;
for (int k = 1; k <= 10; k++)
{
string key = "q" + queryId + "_p" + k;
if (!peptideSection.ContainsKey(key))
{
if (null != lastHit)
{
lastHit.DeltaScore = 1.0;
}
break;
}
string line = peptideSection[key];
if (line == null || line.Equals("-1"))
{
if (null != lastHit)
{
lastHit.DeltaScore = 1.0;
}
break;
}
Match mDetail = this.peptideRegex.Match(line);
if (!mDetail.Success)
{
throw new Exception("Wrong format of peptides : " + line);
}
double score = MyConvert.ToDouble(mDetail.Groups["Score"].Value);
if (score < minScore)
{
if (null != lastHit)
{
lastHit.DeltaScore = 1.0 - score / lastHit.Score;
}
break;
}
bool bSameRank = null != lastHit && score == lastHit.Score;
if (!bSameRank)
{
if (null != lastHit)
{
lastHit.DeltaScore = 1.0 - score / lastHit.Score;
}
rank++;
if (rank > minRank)
{
break;
}
}
IIdentifiedSpectrum mphit;
if (bSameRank)
{
mphit = lastHit;
}
else
{
mphit = new IdentifiedSpectrum();
mphit.IsPrecursorMonoisotopic = isPrecursorMonoisotopic;
mphit.Rank = rank;
mphit.NumMissedCleavages = int.Parse(mDetail.Groups["MissCleavage"].Value);
mphit.TheoreticalMass = MyConvert.ToDouble(mDetail.Groups["TheoreticalMass"].Value);
mphit.ExperimentalMass = queryItem.ExperimentalMass;
mphit.Score = score;
mphit.ExpectValue = ExpectValueCalculator.Calc(mphit.Score, queryItem.MatchCount, 0.05);
mphit.Query.QueryId = queryId;
mphit.Query.ObservedMz = queryItem.Observed;
mphit.Query.Charge = queryItem.Charge;
mphit.Query.MatchCount = queryItem.MatchCount;
if (queryItem.HomologyScore != 0)
{
mphit.Annotations[HomologyScoreKey] = queryItem.HomologyScore;
}
if (CurrentProtease.IsSemiSpecific)
{
mphit.NumProteaseTermini = 1;
}
lastHit = mphit;
}
var pureSeq = mDetail.Groups["Sequence"].Value;
string modification = mDetail.Groups["Modification"].Value;
var seq = ModifySequence(pureSeq, modification);
AssignModification(mphit, modification, CurrentModifications);
string proteins = mDetail.Groups["ProteinNames"].Value;
Match proteinNameMatch = this.proteinNameRegex.Match(proteins);
string key_terms = key + "_terms";
if (!peptideSection.ContainsKey(key_terms))
{
throw new Exception("Mascot version is too old. It's not supported.");
}
string value_terms = peptideSection[key_terms];
Match termsMatch = this.termsRegex.Match(value_terms);
int numProteaseTermini = 0;
while (proteinNameMatch.Success && termsMatch.Success)
{
var fullSeq = MyConvert.Format("{0}.{1}.{2}",
termsMatch.Groups[1].Value,
seq,
termsMatch.Groups[2].Value);
var name = proteinNameMatch.Groups[1].Value.Replace("/", "_");
if (isDecoy)
{
name = DECOY_PREFIX + name;
}
bool findPeptide = false;
for (int i = 0; i < mphit.Peptides.Count; i++)
{
if (mphit.Peptides[i].Sequence == fullSeq)
{
mphit.Peptides[i].AddProtein(name);
findPeptide = true;
break;
}
}
if (!findPeptide)
{
var mp = new IdentifiedPeptide(mphit);
mp.Sequence = fullSeq;
mp.AddProtein(name);
if (CurrentProtease.IsSemiSpecific)
{
int position = Convert.ToInt32(proteinNameMatch.Groups[2].Value);
int count = CurrentProtease.GetNumProteaseTermini(termsMatch.Groups[1].Value[0], pureSeq, termsMatch.Groups[2].Value[0], '-', position);
numProteaseTermini = Math.Max(numProteaseTermini, count);
}
}
proteinNameMatch = proteinNameMatch.NextMatch();
termsMatch = termsMatch.NextMatch();
}
if (CurrentProtease.IsSemiSpecific)
{
mphit.NumProteaseTermini = Math.Max(mphit.NumProteaseTermini, numProteaseTermini);
}
if (!bSameRank)
{
iPeps.Add(mphit);
}
}
string query = "query" + queryId;
Dictionary <string, string> querySection = ParseSection(sr, query);
string title = Uri.UnescapeDataString(querySection["title"]);
SequestFilename sf = this.TitleParser.GetValue(title);
sf.Charge = queryItem.Charge;
if (sf.Experimental == null || sf.Experimental.Length == 0)
{
sf.Experimental = defaultExperimental;
}
foreach (IIdentifiedSpectrum mp in iPeps)
{
mp.Query.Title = title;
mp.Query.FileScan = sf;
}
}
}
return(result);
}
protected Dictionary <int, MascotQueryItem> ParseQueryItems(StreamReader sr, int queryCount, string prefix = "")
{
string line;
SectionClass sc = new SectionClass(prefix + "summary");
while ((line = sr.ReadLine()) != null)
{
if (sc.IsStartLine(line))
{
break;
}
}
var result = new Dictionary <int, MascotQueryItem>();
while ((line = sr.ReadLine()) != null)
{
if (IsSectionEnd(line))
{
break;
}
if (line.StartsWith("qmass"))
{
Match m = queryIdRegex.Match(line);
var item = new MascotQueryItem();
item.QueryId = Convert.ToInt32(m.Groups[1].Value);
item.ExperimentalMass = MyConvert.ToDouble(m.Groups[2].Value);
line = sr.ReadLine();
Match precursor = this.precursorRegex.Match(line);
if (precursor.Success)
{
item.Observed = MyConvert.ToDouble(precursor.Groups[1].Value);
item.Charge = int.Parse(precursor.Groups[2].Value);
}
else
{
var pmr = this.precursorMrRegex.Match(line);
item.Observed = MyConvert.ToDouble(pmr.Groups[1].Value);
item.Charge = 0;
}
line = sr.ReadLine();
if (line.StartsWith("qintensity"))
{
item.Intensity = MyConvert.ToDouble(GetValue(line));
line = sr.ReadLine();
}
if (line.StartsWith("qmatch"))
{
item.MatchCount = Convert.ToInt32(GetValue(line));
line = sr.ReadLine();
}
if (line.StartsWith("qplughole"))
{
item.HomologyScore = Convert.ToDouble(GetValue(line));
}
result[item.QueryId] = item;
}
}
return(result);
}