public static BuildSummaryOptions LoadFromFile(string fileName)
{
BuildSummaryOptions result = new BuildSummaryOptions();
LoadFromFile(fileName, result);
return(result);
}
public IdentifiedSpectrumBuilderResult Build(string parameterFile)
{
Options = new BuildSummaryOptions(parameterFile);
Options.DatasetList.RemoveDisabled();
return(DoBuild(parameterFile));
}
public virtual void SaveDatasetList <T>(BuildSummaryOptions conf) where T : IDatasetOptions, new()
{
Dictionary <string, IDatasetOptions> dsmap = new Dictionary <string, IDatasetOptions>();
foreach (ListViewItem item in this.lvDatFiles.Items)
{
if (item.Selected)
{
var key = string.Empty;
if (item.SubItems.Count >= 3)
{
key = item.SubItems[2].Text;
}
if (!dsmap.ContainsKey(key))
{
var dsoptions = new T();
dsoptions.Name = key;
dsoptions.Parent = conf;
dsmap[key] = dsoptions;
conf.DatasetList.Add(dsoptions);
}
dsmap[key].PathNames.Add(item.SubItems[0].Text);
}
}
}
public IdentifiedSpectrumBuilderResult Build(string parameterFile)
{
Options = new BuildSummaryOptions(parameterFile);
Options.DatasetList.RemoveDisabled();
IIdentifiedProteinBuilder proteinBuilder = new IdentifiedProteinBuilder();
IIdentifiedProteinGroupBuilder groupBuilder = new IdentifiedProteinGroupBuilder();
var fdrCalc = Options.FalseDiscoveryRate.GetFalseDiscoveryRateCalculator();
BuildResult = new DatasetList();
//从配置进行初始化
BuildResult.InitFromOptions(Options.DatasetList, this.Progress, parameterFile);
var totalCount = BuildResult.GetOptimalSpectrumCount();
string optimalResultFile = FileUtils.ChangeExtension(parameterFile, ".optimal");
using (var sw = new StreamWriter(optimalResultFile))
{
new OptimalFileTextWriter().WriteToStream(sw, BuildResult);
UniformProteinFdrOptimalResultCalculator proteinCalc = new UniformProteinFdrOptimalResultCalculator(fdrCalc, Options.GetDecoyGroupFilter())
{
Progress = this.Progress
};
Progress.SetMessage("Filtering PSMs by protein fdr {0}, using peptide fdr {1}...", Options.FalseDiscoveryRate.FdrValue, Options.FalseDiscoveryRate.MaxPeptideFdr);
var groupFilter = Options.FalseDiscoveryRate.FilterOneHitWonder ? new IdentifiedProteinGroupSingleWonderPeptideCountFilter(Options.FalseDiscoveryRate.MinOneHitWonderPeptideCount) : null;
var ret = proteinCalc.GetOptimalResultForGroupFilter(BuildResult, Options.FalseDiscoveryRate.MaxPeptideFdr, Options.FalseDiscoveryRate.FdrValue, groupFilter);
//只保留通过筛选的蛋白质包含的PSMs。
BuildResult.KeepOptimalResultInSetOnly(ret.AcceptedSpectra);
GC.Collect();
GC.WaitForPendingFinalizers();
sw.WriteLine("After SimpleProteinFDR filter {0} with condition {1}, required peptide fdr = {2} ", ret.ProteinFdr, ret.ProteinCondition, ret.PeptideFdr);
BuildResult.ForEach(ds =>
{
sw.WriteLine("Dataset {0}", ds.Options.Name);
OptimalResultConditionUtils.WriteSpectrumBin(sw, ds, f1, f2);
});
//sw.WriteLine();
//new OptimalFileTextWriter().WriteToStream(sw, BuildResult);
return(new IdentifiedSpectrumBuilderResult()
{
Spectra = ret.AcceptedSpectra.ToList(),
PeptideFDR = ret.PeptideFdr,
ProteinFDR = ret.ProteinFdr
});
}
}
public virtual void LoadDatasetList <T>(BuildSummaryOptions options) where T : IDatasetOptions
{
this.lvDatFiles.Items.Clear();
foreach (T dataset in options.DatasetList)
{
foreach (var file in dataset.PathNames)
{
ListViewItem item = this.lvDatFiles.Items.Add(file);
item.SubItems.Add("");
item.SubItems.Add(dataset.Name);
item.Selected = true;
}
}
}
public static void LoadFromFile(string fileName, BuildSummaryOptions result)
{
XElement root = XElement.Load(fileName);
var engineNode = root.Element("SearchEngine");
if (engineNode != null)
{
var engine = engineNode.Value;
SearchEngineType set = EnumUtils.StringToEnum(engine, SearchEngineType.Unknown);
AbstractSummaryConfiguration conf;
switch (set)
{
case SearchEngineType.Unknown:
throw new ArgumentException("Unknown search engine " + engine);
case SearchEngineType.MASCOT:
conf = new MascotDatSummaryConfiguration("");
break;
case SearchEngineType.SEQUEST:
conf = new SequestSummaryConfiguration("");
break;
case SearchEngineType.XTandem:
conf = new XTandemXmlSummaryConfiguration("");
break;
case SearchEngineType.PFind:
conf = new PFindSummaryConfiguration("");
break;
//case SearchEngineType.PEPTIDEPHOPHET:
// op = new PeptideProphetDatasetOptions();
// break;
default:
throw new ArgumentException("It's not defined that how to get SummaryConfiguration for search engine " + engine);
}
conf.LoadFromFile(fileName);
conf.ConvertTo(result);
}
else
{
result.LoadFromFile(fileName);
}
}
public IdentifiedSpectrumBuilderResult Build(string parameterFile)
{
Options = new BuildSummaryOptions(parameterFile);
Options.DatasetList.RemoveDisabled();
IIdentifiedProteinGroupFilter conFilter = Options.Database.GetNotContaminationDescriptionFilter(this.Progress);
var fdrCalc = Options.FalseDiscoveryRate.GetFalseDiscoveryRateCalculator();
BuildResult = new DatasetList();
//从配置进行初始化
BuildResult.InitFromOptions(Options.DatasetList, this.Progress, parameterFile);
var totalCount = BuildResult.GetOptimalSpectrumCount();
string optimalResultFile = FileUtils.ChangeExtension(parameterFile, ".optimal");
using (var sw = new StreamWriter(optimalResultFile))
{
new OptimalFileTextWriter().WriteToStream(sw, BuildResult);
UniformProteinFdrOptimalResultCalculator proteinCalc = new UniformProteinFdrOptimalResultCalculator(fdrCalc, Options.GetDecoyGroupFilter())
{
Progress = this.Progress
};
sw.WriteLine(OptimalFilteredItem.GetHeader());
var uniqueFilter = new IdentifiedProteinGroupUniquePeptideCountFilter(2);
OptimalFilteredItem finalItem = null;
List <IIdentifiedSpectrum> allSpectrum = Options.PeptideRetrieval ? BuildResult.GetSpectra() : null;
int fdrPeptideCount = Options.FalseDiscoveryRate.FdrPeptideCount > 2 ? Options.FalseDiscoveryRate.FdrPeptideCount : 2;
double firstStepFdr = Options.FalseDiscoveryRate.MaxPeptideFdr;
bool bFirst = true;
for (int curPeptideCount = fdrPeptideCount; curPeptideCount >= 2; curPeptideCount--)
{
//重新根据保留的Spectra构建SpectrumBin。
if (!bFirst)
{
BuildResult.BuildSpectrumBin();
}
bFirst = false;
var curItem = new OptimalFilteredItem();
IIdentifiedProteinGroupFilter groupFilter;
bool bNeedFirstStep = curPeptideCount > 2;
if (bNeedFirstStep)
{
Progress.SetMessage("Filtering PSMs by protein fdr {0}, unique peptide count >= 2 and peptide count >= {1} using peptide fdr {2}...", Options.FalseDiscoveryRate.FdrValue, curPeptideCount, firstStepFdr);
//第一步，根据UniquePeptideCount和PeptideCount进行筛选，得到满足蛋白质Fdr要求所对应的肽段fdr。
var countFilter = new IdentifiedProteinGroupPeptideCountFilter(curPeptideCount);
if (conFilter != null)
{
groupFilter = new AndIdentifiedProteinGroupFilter(new IIdentifiedProteinGroupFilter[] { conFilter, uniqueFilter, countFilter });
}
else
{
groupFilter = new AndIdentifiedProteinGroupFilter(new IIdentifiedProteinGroupFilter[] { uniqueFilter, countFilter });
}
curItem.Unique2CountResult = proteinCalc.GetOptimalResultForGroupFilter(BuildResult, firstStepFdr, Options.FalseDiscoveryRate.FdrValue, groupFilter);
firstStepFdr = curItem.Unique2CountResult.PeptideFdr;
//只保留没有被通过筛选的蛋白质包含的PSMs。
BuildResult.KeepOptimalResultInSetOnly(new HashSet <IIdentifiedSpectrum>(curItem.Unique2CountResult.RejectedSpectra));
GC.Collect();
GC.WaitForPendingFinalizers();
}
else
{
curItem.Unique2CountResult = new ProteinFdrFilteredItem();
}
Progress.SetMessage("Filtering PSMs by protein fdr {0}, unique peptide count >= 2 using peptide fdr {1}...", Options.FalseDiscoveryRate.FdrValue, firstStepFdr);
//第二步，根据UniquePeptideCount进行筛选，计算得到满足给定蛋白质fdr的结果。
double secondStepFdr = bNeedFirstStep ? Options.FalseDiscoveryRate.MaxPeptideFdr : firstStepFdr;
if (conFilter != null)
{
groupFilter = new AndIdentifiedProteinGroupFilter(new IIdentifiedProteinGroupFilter[] { conFilter, uniqueFilter });
}
else
{
groupFilter = uniqueFilter;
}
curItem.Unique2Result = proteinCalc.GetOptimalResultForGroupFilter(BuildResult, secondStepFdr, Options.FalseDiscoveryRate.FdrValue, groupFilter);
//只保留没有被通过筛选的蛋白质包含的PSMs。
BuildResult.KeepOptimalResultInSetOnly(new HashSet <IIdentifiedSpectrum>(curItem.Unique2Result.RejectedSpectra));
GC.Collect();
GC.WaitForPendingFinalizers();
curItem.Unique1Result = FilterOneHitWonders(conFilter, proteinCalc);
GC.Collect();
GC.WaitForPendingFinalizers();
sw.WriteLine(curItem.ToString());
if (finalItem == null || finalItem.TotalProteinCount < curItem.TotalProteinCount)
{
finalItem = curItem;
}
curItem = null;
GC.Collect();
GC.WaitForPendingFinalizers();
Console.WriteLine(MyConvert.Format("Filtering PSMs by protein fdr {0}, unique peptide count >= 2 and peptide count >= {1} using peptide fdr {2}...cost {3}.", Options.FalseDiscoveryRate.FdrValue, curPeptideCount, firstStepFdr, SystemUtils.CostMemory()));
}
Progress.SetMessage("Filtering PSMs by protein fdr {0} finished, free memory...", Options.FalseDiscoveryRate.FdrValue);
if (finalItem != null)
{
sw.WriteLine();
sw.WriteLine("Final result : ");
WriteScoreMap(sw, BuildResult, finalItem.Unique2CountResult);
WriteScoreMap(sw, BuildResult, finalItem.Unique2Result);
WriteScoreMap(sw, BuildResult, finalItem.Unique1Result);
var finalSpectra = finalItem.GetSpectra();
if (Options.PeptideRetrieval)
{
Progress.SetMessage("Retrivaling peptides passed maximum peptide FDR for proteins passed protein FDR...");
var proteinBuilder = new IdentifiedProteinBuilder();
var groupBuilder = new IdentifiedProteinGroupBuilder();
List <IIdentifiedProtein> proteins = proteinBuilder.Build(finalSpectra);
List <IIdentifiedProteinGroup> groups = groupBuilder.Build(proteins);
var proteinMap = new Dictionary <string, IIdentifiedProteinGroup>();
foreach (var g in groups)
{
foreach (var p in g)
{
proteinMap[p.Name] = g;
}
}
var savedSpectra = new HashSet <IIdentifiedSpectrum>(finalItem.GetSpectra());
foreach (var spectrum in allSpectrum)
{
if (savedSpectra.Contains(spectrum))
{
continue;
}
var pgs = new HashSet <IIdentifiedProteinGroup>();
foreach (var protein in spectrum.Proteins)
{
IIdentifiedProteinGroup pg;
if (proteinMap.TryGetValue(protein, out pg))
{
pgs.Add(pg);
}
}
//if the spectrum doesn't map to protein passed FDR filter, ignore
//if the spectrum maps to multiple groups, ignore
if (pgs.Count == 0 || pgs.Count > 1)
{
continue;
}
//The spectrum should map to all proteins in the group
if (pgs.First().All(l => spectrum.Proteins.Contains(l.Name)))
{
finalSpectra.Add(spectrum);
}
}
}
BuildResult.ClearSpectra();
GC.Collect();
GC.WaitForPendingFinalizers();
return(new IdentifiedSpectrumBuilderResult()
{
Spectra = finalSpectra,
PeptideFDR = finalItem.Unique2Result.PeptideFdr,
ProteinFDR = Options.FalseDiscoveryRate.FdrValue
});
}
else
{
return(new IdentifiedSpectrumBuilderResult()
{
Spectra = new List <IIdentifiedSpectrum>()
});
}
}
}
