public int GetStudyKey(string SID, FindParam prm, out string patientid, out string studykey)
{
using (new DebugLog())
{
return(sv.GetStudyKey(SID, prm, out patientid, out studykey));
}
}
public int GetStudyList(string SID, FindParam prm, out List <StudyTag> tags, out int count)
{
using (new DebugLog())
{
return(sv.GetStudyList(SID, prm, out tags, out count));
}
}
private void FindMutation(object obj)
{
FindParam param = (FindParam)obj;
for (int i = 0; i < param.PnovoSeqs.Count; i++)
{
if (Progress.IsCancellationPending())
{
return;
}
param.FinishedCount++;
var pnovoseq = param.PnovoSeqs[i];
string source = string.Empty;
int site = -1;
//是否有一个mutation的匹配，不改变酶切位点。
if (FindMutationOneType1(pnovoseq, ref source, ref site))
{
var reference = string.Format("sp|MUL_{0}|type1 source={1} mutation={2}{3}{4}", pnovoseq, source, source[site], site + 1, pnovoseq[site]);
param.Sequences.Add(new Sequence(reference, pnovoseq));
param.Type1Count++;
continue;
}
//是否有一个mutation的匹配，从酶切位点突变为其他氨基酸，导致pnovo解析得到序列更长。
if (FindMutationOneType2(pnovoseq, ref source))
{
param.Sequences.Add(new Sequence("sp|MUL_" + pnovoseq + "|type2 source=" + source, pnovoseq));
param.Type2Count++;
continue;
}
//是否有一个mutation的匹配，从其他氨基酸突变为酶切位点，导致pnovo解析得到序列更短。
if (FindMutationOneType3(pnovoseq, ref source))
{
param.Sequences.Add(new Sequence("sp|MUL_" + pnovoseq + "|type3 source=" + source, pnovoseq));
param.Type3Count++;
continue;
}
}
}
/// <summary>
/// 读取fasta文件，进行数据处理。
/// </summary>
/// <param name="fileName">fasta</param>
/// <returns>result file</returns>
public override IEnumerable <string> Process()
{
HashSet <string> pnovoseqs = new HashSet <string>();
var pnovoParser = new PNovoPlusParser(options.TitleParser);
pnovoParser.Progress = this.Progress;
//找到一个非酶切位点的氨基酸，可代表denovo序列前后氨基酸。
var anotheraa = 'A';
for (int i = 0; i < 26; i++)
{
anotheraa = (char)('A' + i);
if (options.Enzyme.CleaveageResidues.Contains(anotheraa) || options.Enzyme.NotCleaveResidues.Contains(anotheraa))
{
continue;
}
break;
}
Progress.SetRange(0, options.PnovoFiles.Length);
int totalSpectrumCount = 0;
int totalSpectrumPassScore = 0;
foreach (var pnovoFile in options.PnovoFiles)
{
Progress.SetMessage("Reading " + pnovoFile + " ...");
int spectrumCount = pnovoParser.GetSpectrumCount(pnovoFile);
var curSpectra = pnovoParser.ParsePeptides(pnovoFile, 10, options.MinScore);
totalSpectrumCount += spectrumCount;
totalSpectrumPassScore += curSpectra.Count;
RemoveMissCleavagePeptides(anotheraa, curSpectra);
pnovoseqs.UnionWith(from c in curSpectra
from p in c.Peptides
select p.PureSequence);
Progress.Increment(1);
}
var pNovoStat = Path.Combine(options.TargetDirectory, "pNovo.SAP.stat");
using (StreamWriter sw = new StreamWriter(pNovoStat))
{
sw.WriteLine("Total Spectrum Count\t" + totalSpectrumCount.ToString());
sw.WriteLine("Total Peptide-Spectrum-Match Passed Score Filter\t" + totalSpectrumPassScore.ToString());
}
Progress.SetPosition(0);
Progress.SetMessage("Reading " + options.TargetFastaFile + " ...");
var seqs = SequenceUtils.Read(new FastaFormat(), options.TargetFastaFile);
Progress.SetMessage("Digesting sequences ...");
GetDigestPeptide(seqs);
seqs.Clear();
seqs.TrimExcess();
GC.Collect();
GC.WaitForFullGCComplete();
//清除所有跟理论库一样的肽段。
Progress.SetMessage("Removing identical peptides ...");
pnovoseqs.ExceptWith(miss0.Keys);
var pnovoArray = pnovoseqs.ToArray();
pnovoseqs.Clear();
GC.Collect();
GC.WaitForFullGCComplete();
miss0group = miss0.Keys.ToGroupDictionary(m => m.Length);
var type2seqs = new List <Type2Sequence>();
var type2_2 = new List <string>();
foreach (var m in miss1.Keys)
{
int maxpos = -1;
for (int i = 1; i < m.Length; i++)
{
if (options.Enzyme.IsCleavageSite(m[i - 1], m[i], anotheraa))
{
maxpos = i - 1;
break;
}
}
if (maxpos == -1)
{
throw new Exception("There is no misscleavage in " + m);
}
if (maxpos == 0)
{
type2_2.Add(m);
}
else
{
type2seqs.Add(new Type2Sequence()
{
Sequence = m,
PriorSequence = m.Substring(0, maxpos),
PostSequence = m.Substring(maxpos + 1)
});
}
}
miss1type2_1 = type2seqs.ToGroupDictionary(m => GetType2Key(m.Sequence));
miss1type2_2 = type2_2.ToGroupDictionary(m => m.Substring(1));
miss0type3 = miss0.Keys.ToGroupDictionary(m => GetType3Key(m));
type2seqs.Clear();
GC.Collect();
GC.WaitForFullGCComplete();
Progress.SetMessage("Finding mutation ...");
Progress.SetRange(0, pnovoArray.Length);
var pre100 = pnovoArray.Length / 100;
var pre10000 = pnovoArray.Length / 10000;
if (pre10000 == 0)
{
pre10000 = 1;
}
var totalCount = pnovoArray.Length;
var binSize = totalCount / options.ThreadCount;
List <FindParam> fparams = new List <FindParam>();
List <Thread> threads = new List <Thread>();
var startPos = 0;
for (int i = 0; i < options.ThreadCount; i++)
{
int count;
if (i == options.ThreadCount - 1)
{
count = pnovoArray.Length - startPos;
}
else
{
count = binSize;
}
List <string> binSeq = new List <string>();
binSeq.AddRange(pnovoArray.Skip(startPos).Take(count));
startPos = startPos + count;
var aparam = new FindParam()
{
PnovoSeqs = binSeq
};
fparams.Add(aparam);
Thread at = new Thread(this.FindMutation);
threads.Add(at);
at.IsBackground = true;
at.Start(aparam);
}
pnovoArray = null;
GC.Collect();
GC.WaitForFullGCComplete();
var startTime = DateTime.Now;
Progress.SetRange(0, totalCount);
while (true)
{
Thread.Sleep(1000);
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
int finishedCount = fparams.Sum(m => m.FinishedCount);
Progress.SetPosition(finishedCount);
if (finishedCount == 0)
{
continue;
}
var curTime = DateTime.Now;
var costTime = curTime - startTime;
var totalCostTime = new TimeSpan(costTime.Ticks * totalCount / finishedCount);
var finishTime = curTime + new TimeSpan(costTime.Ticks * (totalCount - finishedCount) / finishedCount);
StringBuilder costFormat = new StringBuilder();
if (totalCostTime.TotalHours >= 2.0)
{
costFormat.Append(Math.Truncate(totalCostTime.TotalHours).ToString() + " hours and ");
}
else if (totalCostTime.TotalHours >= 1.0)
{
costFormat.Append("one hour and ");
}
costFormat.Append(totalCostTime.Minutes.ToString() + " minutes");
Progress.SetMessage("Finding mutation {0} / {1}, will cost {2} and finish at {3} ...", finishedCount, totalCount, costFormat, finishTime);
int finishedThreadCount = threads.Count(m => !m.IsAlive);
if (finishedThreadCount == threads.Count)
{
break;
}
}
int type1 = fparams.Sum(m => m.Type1Count);
int type2 = fparams.Sum(m => m.Type2Count);
int type3 = fparams.Sum(m => m.Type3Count);
using (StreamWriter sw = new StreamWriter(pNovoStat, true))
{
sw.WriteLine("Type1 Count\t" + type1.ToString());
sw.WriteLine("Type2 Count\t" + type2.ToString());
sw.WriteLine("Type3 Count\t" + type3.ToString());
}
var singleMutation = (from f in fparams
from s in f.Sequences
select s).ToList();
string newFastaFile = new FileInfo(options.TargetDirectory + "/" + FileUtils.ChangeExtension(new FileInfo(options.DatabaseFastaFile).Name, "mutation.fasta")).FullName;
using (StreamWriter sw = new StreamWriter(newFastaFile))
{
using (StreamReader sr = new StreamReader(options.DatabaseFastaFile))
{
string line = sr.ReadToEnd();
sw.WriteLine(line);
foreach (var seq in singleMutation)
{
sw.WriteLine(">" + seq.Reference);
sw.WriteLine(seq.SeqString);
}
}
}
Progress.SetRange(0, options.PnovoFiles.Length);
var sapSequences = new HashSet <string>(singleMutation.ConvertAll(m => m.SeqString));
List <IIdentifiedSpectrum> allSpectra = new List <IIdentifiedSpectrum>();
foreach (var pnovoFile in options.PnovoFiles)
{
Progress.SetMessage("Reading " + pnovoFile + " ...");
var curSpectra = pnovoParser.ParsePeptides(pnovoFile, 10, options.MinScore);
RemoveMissCleavagePeptides(anotheraa, curSpectra);
curSpectra.RemoveAll(m => !m.Peptides.Any(n => sapSequences.Contains(n.PureSequence)));
allSpectra.AddRange(curSpectra);
Progress.Increment(1);
}
var pNovoPeptides = Path.Combine(options.TargetDirectory, "pNovo.SAP.peptides");
new MascotPeptideTextFormat("\tFileScan\tSequence\tCharge\tScore\tDeltaScore").WriteToFile(pNovoPeptides, allSpectra);
Progress.SetMessage("Finished.");
Progress.End();
return(new string[] { newFastaFile });
}
public int RSKey2StudyKey(string SID, string rskey, out string patid, out string studykey, out string path, out FindParam prm)
{
using (new DebugLog())
{
return(sv.RSKey2StudyKey(SID, rskey, out patid, out studykey, out path, out prm));
}
}
/// <summary>
/// find and replace execute module parameters.
/// </summary>
/// <returns>a StringBuilder that contains a new execute module parameters.</returns>
private StringBuilder FindAndReplaceParameter()
{
Stopwatch stopwatch = Stopwatch.StartNew();
// initialize Dictionaries
Dictionary <int, int> oldExecuteParameters = new Dictionary <int, int>();
Dictionary <int, int> newExecuteParameters = new Dictionary <int, int>();
// convert execute module entry to Dictionary
for (int i = 0; i < FindParam.Split(',').Length; i++)
{
// verify items are numbers
if (int.TryParse(FindParam.Split(',')[i], out int key))
{
if (int.TryParse(FindParam.Split(',')[i + 1], out int value))
{
// add key, value pairs
oldExecuteParameters.Add(key, value);
// skip values as they included in previous step
i++;
}
}
}
// is there any matching parameters?
if (!oldExecuteParameters.Keys.Any(x => ItemsToFind.Contains(x.ToString())))
{
// NO. return old parameters.
NewParameterString = new StringBuilder(MyResources.Strings_DrIPNet_prepend + string.Join(",", FindParam)); //new StringBuilder(FindParam); //
}
else
{
// only one repeating item with one index
List <int> duplicateEntries = ItemsToFind.GroupBy(x => x)
.Where(g => g.Count() > 2)
.Select(y => int.Parse(y.Key)) // new { Element = y.Key, Index = ItemsToFind.IndexOf(y.Key) })
.ToList();
// scan whole Execute parameter entry
foreach (var currentKey in oldExecuteParameters.Keys)
{
// scan whole .csv file for the current key
for (int i = 0; i < ItemsToFind.Count; i++)
{
// string matching would match strings of "4567" and "4567x"
// so match them as numbers
if (int.TryParse(ItemsToFind.ElementAt(i), out int oldKey))
{
// verify both keys are matched.
if (currentKey == oldKey)
{
// initialize the replacement key
int replacementKey = default;
// if the register has multiple entry for each regulator,
if (duplicateEntries.Contains(oldKey))
{
// use the one appropriate to the current regulator
// eg: second for regulator 2
int.TryParse(ItemsToReplace.ElementAt(GetDuplicateIndex(i)), out replacementKey);
}
else
{
// otherwise use corresponding the new register value
// from the excel file.
int.TryParse(ItemsToReplace.ElementAt(i), out replacementKey);
}
// retrieve original value
oldExecuteParameters.TryGetValue(currentKey, out int originalValue);
// modify MB_ActiveProfile instead of using original value
// TODO: This should be aware of if the user entry or a .csv file in use.
newExecuteParameters.Add(replacementKey, ViewModel.ChangeActiveProfile == currentKey ? activeProfile - 1 : originalValue);
// stop scanning and move on to next entry.
break;
}
}
}
}
// return new parameters.
NewParameterString = new StringBuilder(MyResources.Strings_DrIPNet_prepend + string.Join(",", newExecuteParameters.Select(x => x.Key + "," + x.Value)));
}
Debug.WriteLine($"find: {FindParam}\r\nrep: {NewParameterString}");
Debug.WriteLine($"-----------------------------------------------------> total time: {stopwatch.ElapsedMilliseconds}");
// return new StringBuilder
return(NewParameterString);
}
