public static DatasetListOptions LoadOptions(XElement parentNode)
{
var result = new DatasetListOptions();
result.Load(parentNode);
return(result);
}
public BuildSummaryOptions()
{
MergeResult = true;
KeepTopPeptideFromSameEngineButDifferentSearchParameters = true;
SavePeptidesFile = true;
IsSemiPeptideConsiderAsUnique = true;
Database = new DatabaseOptions();
FalseDiscoveryRate = new FalseDiscoveryRateOptions();
Classification = new ClassificationOptions();
PeptideFilter = new PeptideFilterOptions();
DatasetList = new DatasetListOptions();
ConflictType = ResolveSearchEngineConflictTypeFactory.QValue;
MinimumEngineAgreeCount = 1;
PeptideRetrieval = false;
}
public void InitFromOptions(DatasetListOptions dsOptions, IProgressCallback progress, string paramFile)
{
this.Clear();
this.conflictFunc = dsOptions.Options.GetConflictFunc();
this.fdrCalc = dsOptions.Options.FalseDiscoveryRate.GetFalseDiscoveryRateCalculator();
IFilter <IIdentifiedSpectrum> decoyFilter = null;
if (dsOptions.Options.FalseDiscoveryRate.FilterByFdr)
{
decoyFilter = dsOptions.Options.GetDecoySpectrumFilter();
}
this.dsOptions = dsOptions;
long afterFirstMemory = 0;
DateTime afterFirstTime = DateTime.Now;
var totalCount = dsOptions.Sum(l => l.PathNames.Count);
var usedCount = 0;
for (int i = 0; i < dsOptions.Count; i++)
{
var m = dsOptions[i];
var builder = m.GetBuilder();
builder.Progress = progress;
Dataset ds = new Dataset(m);
//首先，获取所有通过了固定筛选标准的谱图。
ds.Spectra = builder.ParseFromSearchResult();
ds.PSMPassedFixedCriteriaCount = ds.Spectra.Count;
if (dsOptions.Options.FalseDiscoveryRate.FilterByFdr)
{
//对每个谱图设置是否来自诱饵库
progress.SetMessage("Assigning decoy information...");
DecoyPeptideBuilder.AssignDecoy(ds.Spectra, decoyFilter);
var decoyCount = ds.Spectra.Count(l => l.FromDecoy);
if (decoyCount == 0)
{
throw new Exception(string.Format("No decoy protein found at dataset {0}, make sure the protein access number parser and the decoy pattern are correctly defined!", m.Name));
}
progress.SetMessage("{0} decoys out of {1} hits found", decoyCount, ds.Spectra.Count);
ds.BuildSpectrumBin();
ds.CalculateCurrentFdr();
ds.PushCurrentOptimalResults(string.Format("Before maximum peptide fdr {0}", dsOptions.Options.FalseDiscoveryRate.MaxPeptideFdr));
progress.SetMessage("Filtering by maximum peptide fdr {0} ...", dsOptions.Options.FalseDiscoveryRate.MaxPeptideFdr);
ds.FilterByFdr(dsOptions.Options.FalseDiscoveryRate.MaxPeptideFdr);
ds.Spectra = ds.GetUnconflictedOptimalSpectra();
ds.BuildSpectrumBin();
ds.CalculateCurrentFdr();
ds.PushCurrentOptimalResults(string.Format("After maximum peptide fdr {0}", dsOptions.Options.FalseDiscoveryRate.MaxPeptideFdr));
}
this.Add(ds);
if (i == 0)
{
afterFirstMemory = Process.GetCurrentProcess().WorkingSet64 / (1024 * 1024);
afterFirstTime = DateTime.Now;
}
else
{
usedCount += m.PathNames.Count;
long currMemory = Process.GetCurrentProcess().WorkingSet64 / (1024 * 1024);
double averageCost = (double)(currMemory - afterFirstMemory) / usedCount;
double estimatedCost = afterFirstMemory + averageCost * totalCount;
DateTime currTime = DateTime.Now;
var averageTime = currTime.Subtract(afterFirstTime).TotalMinutes / usedCount;
var finishTime = afterFirstTime.AddMinutes(averageTime * (totalCount - dsOptions[0].PathNames.Count));
progress.SetMessage("{0}/{1} datasets, cost {2}M, avg {3:0.0}M, need {4:0.0}M, will finish at {5:MM-dd HH:mm:ss}", (i + 1), dsOptions.Count, currMemory, averageCost, estimatedCost, finishTime);
}
}
//初始化实验列表
this.ForEach(m => m.InitExperimentals());
if (dsOptions.Count > 1)
{
if (dsOptions.Options.KeepTopPeptideFromSameEngineButDifferentSearchParameters)
{
//合并/删除那些相同搜索引擎，不同参数得到的结果。
ProcessDatasetFromSameEngine(progress, (peptides, score) => IdentifiedSpectrumUtils.KeepTopPeptideFromSameEngineDifferentParameters(peptides, score), false);
}
else
{
ProcessDatasetFromSameEngine(progress, (peptides, score) => IdentifiedSpectrumUtils.KeepUnconflictPeptidesFromSameEngineDifferentParameters(peptides, score), true);
}
//初始化不同搜索引擎搜索的dataset之间的overlap关系。
this.OverlapBySearchEngine = FindOverlap((m1, m2) => m1.Options.SearchEngine != m2.Options.SearchEngine);
//初始化没有交集的dataset
var overlaps = new HashSet <Dataset>(from m in OverlapBySearchEngine
from s in m
select s);
this.NoOverlaps = this.Where(m => !overlaps.Contains(m)).ToList();
if (OverlapBySearchEngine.Count > 0 && dsOptions.Options.FalseDiscoveryRate.FilterByFdr)
{
//根据最大的fdr进行筛选。
progress.SetMessage("Filtering PSMs by maximum fdr {0}, considering multiple engine overlap...", dsOptions.Options.FalseDiscoveryRate.MaxPeptideFdr);
var realFdr = this.FilterByFdr(dsOptions.Options.FalseDiscoveryRate.MaxPeptideFdr);
if (realFdr.ConflictSpectra.Count > 0)
{
new MascotPeptideTextFormat(UniformHeader.PEPTIDE_HEADER).WriteToFile(Path.ChangeExtension(paramFile, ".conflicted.peps"), realFdr.ConflictSpectra);
}
//保留每个dataset的spectra为筛选后的结果，以用于后面的迭代。
this.ForEach(m =>
{
m.Spectra = m.GetUnconflictedOptimalSpectra();
});
}
}
else
{
this.NoOverlaps = new List <Dataset>(this);
this.OverlapBySearchEngine = new List <List <Dataset> >();
}
}
