public ScanBasedTagSearchEngine(
            LcMsRun run,
            ISequenceTagFinder seqTagFinder,
            LcMsPeakMatrix featureFinder,
            FastaDatabase fastaDb,
            Tolerance tolerance,
            AminoAcidSet aaSet,
            CompositeScorerFactory ms2ScorerFactory = null,
            int minMatchedTagLength = DefaultMinMatchedTagLength,
            double maxSequenceMass = 50000.0,
            int minProductIonCharge = 1,
            int maxProductIonCharge = 20)
        {
            _run = run;
            _featureFinder = featureFinder;
            
            _searchableDb = new SearchableDatabase(fastaDb);

            _tolerance = tolerance;
            _aaSet = aaSet;
            _minMatchedTagLength = minMatchedTagLength;
            _maxSequenceMass = maxSequenceMass;
            _minProductIonCharge = minProductIonCharge;
            _maxProductIonCharge = maxProductIonCharge;
            MinScan = int.MinValue;
            MaxScan = int.MaxValue;
            _ms2ScorerFactory = ms2ScorerFactory;
            _seqTagFinder = seqTagFinder;
        }
Exemple #2
0
        public bool RunSearch(double corrThreshold = 0.7, CancellationToken? cancellationToken = null, IProgress<ProgressData> progress = null)
        {
            // Get the Normalized spec file/folder path
            SpecFilePath = MassSpecDataReaderFactory.NormalizeDatasetPath(SpecFilePath);

            var prog = new Progress<ProgressData>();
            var progData = new ProgressData(progress);
            if (progress != null)
            {
                prog = new Progress<ProgressData>(p =>
                {
                    progData.Status = p.Status;
                    progData.StatusInternal = p.StatusInternal;
                    progData.Report(p.Percent);
                });
            }

            var sw = new Stopwatch();
            var swAll = new Stopwatch();
            swAll.Start();
            ErrorMessage = string.Empty;

            Console.Write(@"Reading raw file...");
            progData.Status = "Reading spectra file";
            progData.StepRange(10.0);
            sw.Start();

            _run = PbfLcMsRun.GetLcMsRun(SpecFilePath, 0, 0, prog);

            _ms2ScanNums = _run.GetScanNumbers(2).ToArray();
            _isolationWindowTargetMz = new double[_run.MaxLcScan + 1];
            foreach (var ms2Scan in _ms2ScanNums)
            {
                var ms2Spec = _run.GetSpectrum(ms2Scan) as ProductSpectrum;
                if (ms2Spec == null) continue;
                _isolationWindowTargetMz[ms2Scan] = ms2Spec.IsolationWindow.IsolationWindowTargetMz;
            }

            
            sw.Stop();
            Console.WriteLine(@"Elapsed Time: {0:f1} sec", sw.Elapsed.TotalSeconds);

            progData.StepRange(20.0);
            ISequenceFilter ms1Filter;
            if (this.ScanNumbers != null && this.ScanNumbers.Any())
            {
                ms1Filter = new SelectedMsMsFilter(this.ScanNumbers);
            }
            else if (string.IsNullOrWhiteSpace(FeatureFilePath))
            {
                // Checks whether SpecFileName.ms1ft exists
                var ms1FtFilePath = MassSpecDataReaderFactory.ChangeExtension(SpecFilePath, LcMsFeatureFinderLauncher.FileExtension);
                if (!File.Exists(ms1FtFilePath))
                {
                    Console.WriteLine(@"Running ProMex...");
                    sw.Start();
                    var param = new LcMsFeatureFinderInputParameter
                    {
                        InputPath = SpecFilePath,
                        MinSearchMass = MinSequenceMass,
                        MaxSearchMass = MaxSequenceMass,
                        MinSearchCharge = MinPrecursorIonCharge,
                        MaxSearchCharge = MaxPrecursorIonCharge,
                        CsvOutput = false,
                        ScoreReport = false,
                        LikelihoodScoreThreshold = -10
                    };
                    var featureFinder = new LcMsFeatureFinderLauncher(param);
                    featureFinder.Run();
                }
                sw.Reset();
                sw.Start();
                Console.Write(@"Reading ProMex results...");
                ms1Filter = new Ms1FtFilter(_run, PrecursorIonTolerance, ms1FtFilePath, -10);
            }
            else
            {
                sw.Reset();
                sw.Start();
                var extension = Path.GetExtension(FeatureFilePath);
                if (extension.ToLower().Equals(".csv"))
                {
                    Console.Write(@"Reading ICR2LS/Decon2LS results...");
                    ms1Filter = new IsosFilter(_run, PrecursorIonTolerance, FeatureFilePath);
                }
                else if (extension.ToLower().Equals(".ms1ft"))
                {
                    Console.Write(@"Reading ProMex results...");
                    ms1Filter = new Ms1FtFilter(_run, PrecursorIonTolerance, FeatureFilePath, -10);
                }
                else if (extension.ToLower().Equals(".msalign"))
                {
                    Console.Write(@"Reading MS-Align+ results...");
                    ms1Filter = new MsDeconvFilter(_run, PrecursorIonTolerance, FeatureFilePath);
                }
                else ms1Filter = null; //new Ms1FeatureMatrix(_run);
            }

            sw.Stop();
            Console.WriteLine(@"Elapsed Time: {0:f1} sec", sw.Elapsed.TotalSeconds);

            
            // pre-generate deconvoluted spectra for scoring
            _massBinComparer = new FilteredProteinMassBinning(AminoAcidSet, MaxSequenceMass+1000);

            _ms2ScorerFactory2 = new CompositeScorerFactory(_run, _massBinComparer, AminoAcidSet,
                                                               MinProductIonCharge, MaxProductIonCharge, ProductIonTolerance);
            sw.Reset();
            Console.WriteLine(@"Generating deconvoluted spectra for MS/MS spectra...");
            sw.Start();
            var pfeOptions = new ParallelOptions
            {
                MaxDegreeOfParallelism = MaxNumThreads,
                CancellationToken = cancellationToken ?? CancellationToken.None
            };
            Parallel.ForEach(_ms2ScanNums, pfeOptions, ms2ScanNum =>
            {
                _ms2ScorerFactory2.DeconvonluteProductSpectrum(ms2ScanNum);
            });
            sw.Stop();
            Console.WriteLine(@"Elapsed Time: {0:f1} sec", sw.Elapsed.TotalSeconds);

            progData.StepRange(10.0);
            progData.Status = "Reading Fasta File";

            // Target database
            var targetDb = new FastaDatabase(DatabaseFilePath);
            targetDb.Read();
            
            // Generate sequence tags for all MS/MS spectra
            if (TagBasedSearch)
            {
                progData.StepRange(25.0);
                progData.Status = "Generating Sequence Tags";
                sw.Reset();
                Console.WriteLine(@"Generating sequence tags for MS/MS spectra...");
                sw.Start();
                var seqTagGen = GetSequenceTagGenerator();
                _tagMs2ScanNum = seqTagGen.GetMs2ScanNumsContainingTags().ToArray();
                sw.Stop();
                Console.WriteLine(@"Elapsed Time: {0:f1} sec", sw.Elapsed.TotalSeconds);
                _tagSearchEngine = new ScanBasedTagSearchEngine(_run, seqTagGen, new LcMsPeakMatrix(_run, ms1Filter), targetDb, ProductIonTolerance, AminoAcidSet,
                                _ms2ScorerFactory2,
                                ScanBasedTagSearchEngine.DefaultMinMatchedTagLength,
                                MaxSequenceMass, MinProductIonCharge, MaxProductIonCharge);                
            }
            
            var specFileName = MassSpecDataReaderFactory.RemoveExtension(Path.GetFileName(SpecFilePath));
            var targetOutputFilePath = Path.Combine(OutputDir, specFileName + TargetFileNameEnding);
            var decoyOutputFilePath = Path.Combine(OutputDir, specFileName + DecoyFileNameEnding);
            var tdaOutputFilePath = Path.Combine(OutputDir, specFileName + TdaFileNameEnding);

            progData.StepRange(60.0);
            progData.Status = "Running Target search";

            if (RunTargetDecoyAnalysis.HasFlag(DatabaseSearchMode.Target))
            {
                sw.Reset();
                Console.Write(@"Reading the target database...");
                sw.Start();
                targetDb.Read();
                sw.Stop();
                Console.WriteLine(@"Elapsed Time: {0:f1} sec", sw.Elapsed.TotalSeconds);

                var targetMatches = new SortedSet<DatabaseSequenceSpectrumMatch>[_run.MaxLcScan + 1];
                
                progData.MaxPercentage = 42.5;
                if (TagBasedSearch)
                {
                    sw.Reset();
                    Console.WriteLine(@"Tag-based searching the target database");
                    sw.Start();
                    RunTagBasedSearch(targetMatches, targetDb, null, prog);
                    Console.WriteLine(@"Target database tag-based search elapsed Time: {0:f1} sec", sw.Elapsed.TotalSeconds);
                }
                progData.MaxPercentage = 60.0;

                sw.Reset();
                Console.WriteLine(@"Searching the target database");
                sw.Start();
                RunSearch(targetMatches, targetDb, ms1Filter, null, prog);
                Console.WriteLine(@"Target database search elapsed Time: {0:f1} sec", sw.Elapsed.TotalSeconds);

                // calculate spectral e-value usign generating function
                sw.Reset();
                Console.WriteLine(@"Calculating spectral E-values for target-spectrum matches");
                sw.Start();
                var bestTargetMatches = RunGeneratingFunction(targetMatches);
                WriteResultsToFile(bestTargetMatches, targetOutputFilePath, targetDb);
                sw.Stop();
                Console.WriteLine(@"Target-spectrum match E-value calculation elapsed Time: {0:f1} sec", sw.Elapsed.TotalSeconds);
            }

            progData.StepRange(95.0); // total to 95%
            progData.Status = "Running Decoy search";

            if (RunTargetDecoyAnalysis.HasFlag(DatabaseSearchMode.Decoy))
            {
                // Decoy database
                sw.Reset();
                sw.Start();
                var decoyDb = targetDb.Decoy(null, true);

                Console.Write(@"Reading the decoy database...");
                decoyDb.Read();
                Console.WriteLine(@"Elapsed Time: {0:f1} sec", sw.Elapsed.TotalSeconds);

                progData.MaxPercentage = 77.5;
                var decoyMatches = new SortedSet<DatabaseSequenceSpectrumMatch>[_run.MaxLcScan + 1];
                if (TagBasedSearch)
                {
                    sw.Reset();
                    Console.WriteLine(@"Tag-based searching the decoy database");
                    sw.Start();
                    RunTagBasedSearch(decoyMatches, decoyDb, null, prog);
                    Console.WriteLine(@"Decoy database tag-based search elapsed Time: {0:f1} sec", sw.Elapsed.TotalSeconds);                    
                }
                progData.MaxPercentage = 95.0;

                sw.Reset();
                Console.WriteLine(@"Searching the decoy database");
                sw.Start();
                RunSearch(decoyMatches, decoyDb, ms1Filter, null, prog);
                Console.WriteLine(@"Decoy database search elapsed Time: {0:f1} sec", sw.Elapsed.TotalSeconds);

                // calculate spectral e-value usign generating function
                sw.Reset();
                Console.WriteLine(@"Calculating spectral E-values for decoy-spectrum matches");
                sw.Start();
                var bestDecoyMatches = RunGeneratingFunction(decoyMatches);
                WriteResultsToFile(bestDecoyMatches, decoyOutputFilePath, decoyDb);
                sw.Stop();
                Console.WriteLine(@"Decoy-spectrum match E-value calculation elapsed Time: {0:f1} sec", sw.Elapsed.TotalSeconds);
            }

            progData.StepRange(100.0);
            progData.Status = "Writing combined results file";
            if (RunTargetDecoyAnalysis.HasFlag(DatabaseSearchMode.Both))
            {
                // Add "Qvalue" and "PepQValue"
                var fdrCalculator = new FdrCalculator(targetOutputFilePath, decoyOutputFilePath);
                if (fdrCalculator.HasError())
                {
                    ErrorMessage = fdrCalculator.ErrorMessage;
                    Console.WriteLine(@"Error computing FDR: " + fdrCalculator.ErrorMessage);
                    return false;
                }

                fdrCalculator.WriteTo(tdaOutputFilePath);
            }
            progData.Report(100.0);

            Console.WriteLine(@"Done.");
            swAll.Stop();
            Console.WriteLine(@"Total elapsed time for search: {0:f1} sec ({1:f2} min)", swAll.Elapsed.TotalSeconds, swAll.Elapsed.TotalMinutes);

            return true;
        }