Added support for bruker TOF files, still buggy

Author: Alexander-Sol

MetaMorpheus/CMD/CommandLineSettings.cs

@@ -99,17 +99,17 @@ public void ValidateCommandLineSettings()
 
             Spectra.AddRange(spectraFromDirectories);
 
-            // Correct .tdf/.tdf_bin file paths to their parent .d directory
-            // This handles the case where a user provides a path to a .tdf file directly
+            // Correct Bruker inner-file paths (.tdf/.tdf_bin for timsTOF, .baf for regular TOF) to their parent .d directory
+            // This handles the case where a user provides a path to one of these files directly
             for (int i = 0; i < Spectra.Count; i++)
             {
                 string ext = Path.GetExtension(Spectra[i]).ToLowerInvariant();
-                if (ext == ".tdf" || ext == ".tdf_bin")
+                if (ext == ".tdf" || ext == ".tdf_bin" || ext == ".baf")
                 {
                     string parentDir = Path.GetDirectoryName(Spectra[i]);
-                    if (parentDir != null && 
-                        parentDir.EndsWith(".d", StringComparison.OrdinalIgnoreCase) && 
-                        IsValidTimsTofDirectory(parentDir))
+                    if (parentDir != null &&
+                        parentDir.EndsWith(".d", StringComparison.OrdinalIgnoreCase) &&
+                        IsValidBrukerDirectory(parentDir))
                     {
                         Spectra[i] = parentDir;
                     }
@@ -120,7 +120,7 @@ public void ValidateCommandLineSettings()
             Spectra = Spectra.Distinct().ToList();
 
             // remove spectra directories, after their spectra files have been added
-            // but keep .d directories as they are valid timsTOF spectra folders
+            // but keep .d directories as they are valid Bruker spectra folders
             Spectra.RemoveAll(p => Directory.Exists(p) && !p.EndsWith(".d", StringComparison.OrdinalIgnoreCase));
 
             IEnumerable<string> fileNames = Tasks.Concat(Databases).Concat(Spectra);
@@ -209,12 +209,13 @@ private static void FindSpectraFilesRecursive(string path, List<string> spectraF
             {
                 if (GlobalVariables.AcceptedSpectraFormats.Contains(GlobalVariables.GetFileExtension(path).ToLowerInvariant()))
                 {
-                    // If a .tdf or .tdf_bin file is found, check if the parent directory is a valid .d folder and add that instead
-                    if (new[] { ".tdf", ".tdf_bin" }.Contains(GlobalVariables.GetFileExtension(path).ToLowerInvariant()))
+                    // If a Bruker inner file is found (.tdf/.tdf_bin for timsTOF, .baf for regular TOF),
+                    // check if the parent directory is a valid .d folder and add that instead
+                    if (new[] { ".tdf", ".tdf_bin", ".baf" }.Contains(GlobalVariables.GetFileExtension(path).ToLowerInvariant()))
                     {
                         var parentDirectory = Directory.GetParent(path);
-                        if (GlobalVariables.GetFileExtension(parentDirectory.FullName).ToLowerInvariant() == ".d" && IsValidTimsTofDirectory(parentDirectory.FullName))
-                            path = parentDirectory.FullName; // add the .d folder instead of the individual tdf/tsf file
+                        if (GlobalVariables.GetFileExtension(parentDirectory.FullName).ToLowerInvariant() == ".d" && IsValidBrukerDirectory(parentDirectory.FullName))
+                            path = parentDirectory.FullName; // add the .d folder instead of the individual inner file
                     }
 
                     spectraFiles.Add(path);
@@ -232,8 +233,8 @@ private static void FindSpectraFilesRecursive(string path, List<string> spectraF
                 // Check if this is a .d folder
                 if (path.EndsWith(".d", StringComparison.OrdinalIgnoreCase))
                 {
-                    // Check if it's a valid timsTOF data folder
-                    if (IsValidTimsTofDirectory(path))
+                    // Check if it's a valid Bruker data folder (timsTOF or regular TOF)
+                    if (IsValidBrukerDirectory(path))
                     {
                         spectraFiles.Add(path);
 
@@ -255,17 +256,24 @@ private static void FindSpectraFilesRecursive(string path, List<string> spectraF
         }
 
         /// <summary>
-        /// Checks if a .d directory is a valid timsTOF data folder.
+        /// Checks if a .d directory is a valid Bruker data folder that MetaMorpheus can read,
+        /// i.e. a timsTOF folder (analysis.tdf + analysis.tdf_bin) or a regular TOF folder (analysis.baf).
         /// </summary>
-        private static bool IsValidTimsTofDirectory(string directoryPath)
+        private static bool IsValidBrukerDirectory(string directoryPath)
         {
-            // for data with tims enabled, we need to have both the .tdf and .tdf_bin files
+            // timsTOF data: needs both the .tdf and .tdf_bin files
             if (File.Exists(Path.Combine(directoryPath, "analysis.tdf")) &&
                 File.Exists(Path.Combine(directoryPath, "analysis.tdf_bin")))
             {
                 return true;
             }
 
+            // regular TOF data: has a .baf file
+            if (File.Exists(Path.Combine(directoryPath, "analysis.baf")))
+            {
+                return true;
+            }
+
             return false;
         }
     }

MetaMorpheus/EngineLayer/GlobalVariables.cs

@@ -76,7 +76,7 @@ public static class GlobalVariables
         public static void SetUpGlobalVariables()
         {
             AcceptedDatabaseFormats = new List<string> { ".fasta", ".fa", ".xml", ".msp" };
-            AcceptedSpectraFormats = new List<string> { ".raw", ".mzml", ".mgf", ".msalign", ".tdf", ".tdf_bin", ".d" };
+            AcceptedSpectraFormats = new List<string> { ".raw", ".mzml", ".mgf", ".msalign", ".tdf", ".tdf_bin", ".baf", ".d" };
             AnalyteType = AnalyteType.Peptide;
             _InvalidAminoAcids = new char[] { 'X', 'B', 'J', 'Z', ':', '|', ';', '[', ']', '{', '}', '(', ')', '+', '-' };
             ExperimentalDesignFileName = "ExperimentalDesign.tsv";

MetaMorpheus/GUI/MainWindow.xaml.cs

@@ -512,7 +512,7 @@ private void Window_Drop(object sender, DragEventArgs e)
         /// </summary>
         private void AddSpectraFile_Click(object sender, RoutedEventArgs e)
         {
-            var openPicker = StartOpenFileDialog("Spectra Files(*.raw;*.mzML;*.mgf;*ms2.msalign;*.tdf;*.tdf_bin)|*.raw;*.mzML;*.mgf;*ms2.msalign;*.tdf;*.tdf_bin");
+            var openPicker = StartOpenFileDialog("Spectra Files(*.raw;*.mzML;*.mgf;*ms2.msalign;*.tdf;*.tdf_bin;*.baf)|*.raw;*.mzML;*.mgf;*ms2.msalign;*.tdf;*.tdf_bin;*.baf");
 
             if (openPicker.ShowDialog() == true)
             {
@@ -1679,25 +1679,25 @@ private void AddPreRunFileRecursiveHelper(string path)
             }  
             if (Directory.Exists(path))
             {
-                // .d folders are usually timsTOF data - if it's a valid timsTOF file, don't recurse into it
-                if (path.EndsWith(".d", StringComparison.OrdinalIgnoreCase) 
-                    && AddOrWarnTimsTofDirectory(path))
+                // .d folders are Bruker data - if it's a valid Bruker data file, don't recurse into it
+                if (path.EndsWith(".d", StringComparison.OrdinalIgnoreCase)
+                    && AddOrWarnBrukerDirectory(path))
                     return; // base case 2
                 foreach (var entry in Directory.EnumerateFileSystemEntries(path))
                     AddPreRunFileRecursiveHelper(entry);
             }
         }
 
-        private bool AddOrWarnTimsTofDirectory(string directoryPath)
+        private bool AddOrWarnBrukerDirectory(string directoryPath)
         {
-            if (IsValidTimsTofDirectory(directoryPath))
+            if (IsValidBrukerDirectory(directoryPath))
             {
                 AddPreRunFile(directoryPath);
                 return true;
             }
             else
             {
-                NotificationHandler(null, new StringEventArgs($"{directoryPath} is not a valid timsTOF data file.", null));
+                NotificationHandler(null, new StringEventArgs($"{directoryPath} is not a valid Bruker data file.", null));
             }
             return false;
         }
@@ -1716,6 +1716,27 @@ public bool IsValidTimsTofDirectory(string directoryPath)
                 && File.Exists(Path.Combine(directoryPath, "analysis.tdf_bin"));
         }
 
+        /// <summary>
+        /// Determines whether the specified directory is a valid regular (non-tims) Bruker TOF data directory.
+        /// </summary>
+        /// <param name="directoryPath">The full path to the directory to validate. The path must end with the ".d" extension.</param>
+        /// <returns>true if the directory exists, ends with ".d", and contains an "analysis.baf" file; otherwise, false.</returns>
+        public bool IsValidBafDirectory(string directoryPath)
+        {
+            return directoryPath.EndsWith(".d", StringComparison.OrdinalIgnoreCase)
+                && Directory.Exists(directoryPath)
+                && File.Exists(Path.Combine(directoryPath, "analysis.baf"));
+        }
+
+        /// <summary>
+        /// Determines whether the specified .d directory is a Bruker data directory that MetaMorpheus can read,
+        /// i.e. either a timsTOF directory or a regular TOF directory.
+        /// </summary>
+        public bool IsValidBrukerDirectory(string directoryPath)
+        {
+            return IsValidTimsTofDirectory(directoryPath) || IsValidBafDirectory(directoryPath);
+        }
+
         private void AddPreRunFile(string filePath)
         {
             if (!RunTasksButton.IsEnabled)
@@ -1794,7 +1815,7 @@ private void AddPreRunFile(string filePath)
                     break;
                 case ".tdf":
                 case ".tdf_bin":
-                    // for Bruker timsTof files, the .tdf and .tdf_bin files are in a ".d" directory 
+                    // for Bruker timsTof files, the .tdf and .tdf_bin files are in a ".d" directory
                     // the fileReader is designed to take the path to the .d directory instead of the individual files
                     var parent = Path.GetDirectoryName(filePath);
                     if (!IsValidTimsTofDirectory(parent))
@@ -1804,6 +1825,17 @@ private void AddPreRunFile(string filePath)
                     }
                     filePath = parent; // change the file path to the .d directory so that it can be properly read by the file reader
                     goto case ".d";
+                case ".baf":
+                    // for regular Bruker TOF files, the analysis.baf file is in a ".d" directory
+                    // the fileReader is designed to take the path to the .d directory instead of the individual files
+                    var bafParent = Path.GetDirectoryName(filePath);
+                    if (!IsValidBafDirectory(bafParent))
+                    {
+                        NotificationHandler(null, new StringEventArgs($"{bafParent} is not a valid Bruker data file; {filePath} could not be added.", null));
+                        return;
+                    }
+                    filePath = bafParent; // change the file path to the .d directory so that it can be properly read by the file reader
+                    goto case ".d";
                 case ".d": // Bruker data files are directories that contain .d files
                     NotificationHandler(null, new StringEventArgs("Quantification and calibration are not currently supported for Bruker data files. All other features of MetaMorpheus will function.", null));
                     goto case ".mzml";

MetaMorpheus/GUI/MetaDraw/MetaDraw.xaml.cs

@@ -110,8 +110,9 @@ private void AddFile(string filePath)
 
             if (GlobalVariables.AcceptedSpectraFormats.Contains(theExtension))
             {
-                // If a bruker timsTof file was selected, we actually want the parent folder
-                if(theExtension == ".tdf" || theExtension == ".tdf_bin")
+                // If a Bruker inner file was selected (.tdf/.tdf_bin for timsTOF, .baf for regular TOF),
+                // we actually want the parent .d folder
+                if(theExtension == ".tdf" || theExtension == ".tdf_bin" || theExtension == ".baf")
                 {
                     filePath = Path.GetDirectoryName(filePath);
                 }

MetaMorpheus/MetaMorpheusSetup/Product.wxs

@@ -360,6 +360,14 @@
       <Component Id="src_timsdata.dll" Guid="22BAA8D2-623B-493E-932E-666A27ECC5A0">
       	<File Id="src_timsdata.dll" Name="timsdata.dll" Source="$(var.GUI_TargetDir)timsdata.dll" />
       </Component>
+      <!--Native Bruker baf2sql library (+ its vcomp110 dependency) for reading regular (non-tims)
+          Bruker TOF .d files. mzLib P/Invokes baf2sql_c by bare name, so it loads from the app root.-->
+      <Component Id="src_baf2sql_c.dll" Guid="ebbca9cb-0e89-4ce8-b540-04bd0d327da4">
+      	<File Id="src_baf2sql_c.dll" Name="baf2sql_c.dll" Source="$(var.GUI_TargetDir)baf2sql_c.dll" />
+      </Component>
+      <Component Id="src_vcomp110.dll" Guid="e18833e2-6562-43bf-9001-d148aa9da919">
+      	<File Id="src_vcomp110.dll" Name="vcomp110.dll" Source="$(var.GUI_TargetDir)vcomp110.dll" />
+      </Component>
       <Component Id="src_ThermoFisher.CommonCore.BackgroundSubtraction.dll" Guid="F3DA60DB-CF5F-4671-9F3C-1033780261DD">
         <File Id="src_ThermoFisher.CommonCore.BackgroundSubtraction.dll" Name="ThermoFisher.CommonCore.BackgroundSubtraction.dll" Source="$(var.GUI_TargetDir)ThermoFisher.CommonCore.BackgroundSubtraction.dll" />
       </Component>

MetaMorpheus/Test/MsDataFileTest.cs

@@ -19,9 +19,37 @@ public static void Setup()
             Environment.CurrentDirectory = TestContext.CurrentContext.TestDirectory;
         }
 
+        /// <summary>
+        /// Verifies that a regular (non-tims) centroided Bruker TOF .d folder (identified by
+        /// analysis.baf) loads through MetaMorpheus's file manager and exposes the expected spectra.
+        /// </summary>
+        /// <remarks>
+        /// Only centroid data is covered here. Profile-mode Bruker TOF data currently throws
+        /// IndexOutOfRangeException in mzLib's BrukerFileReader.GetSpectraData when peak filtering
+        /// is applied (which MyFileManager always does); add profile coverage once that mzLib bug is fixed.
+        /// </remarks>
+        [Test]
+        public static void TestBrukerTofFileLoadsThroughMyFileManager()
+        {
+            string dotDPath = Path.Combine(TestContext.CurrentContext.TestDirectory, @"TestData\centroid_1x_MS1_4x_autoMS2.d");
+            Assert.That(Directory.Exists(dotDPath), Is.True, "Test requires the centroid Bruker TOF .d folder to exist");
+
+            var dataFile = new MyFileManager(true).LoadFile(dotDPath, new CommonParameters());
+
+            Assert.That(dataFile.NumSpectra, Is.EqualTo(5));
+            var secondScan = dataFile.Scans[1];
+            Assert.That(secondScan.MsnOrder, Is.EqualTo(2));
+            Assert.That(secondScan.Polarity, Is.EqualTo(Polarity.Positive));
+            Assert.That(secondScan.DissociationType, Is.EqualTo(DissociationType.CID));
+            Assert.That(secondScan.IsCentroid, Is.True);
+            Assert.That(secondScan.NativeId, Is.EqualTo("scan=2"));
+            Assert.That(secondScan.SelectedIonMZ, Is.EqualTo(721.86865).Within(0.001));
+        }
+
         [Test]
         [TestCase(@"TestData\ok.mgf", @"TestData\okk.xml")]
         [TestCase(@"TestData\snippet.d", @"TestData\gapdh.fasta")]
+        [TestCase(@"TestData\centroid_1x_MS1_4x_autoMS2.d", @"TestData\gapdh.fasta")]
         [TestCase(@"TopDownTestData\JurkatTopDownRep2Fract1_ms2.msalign", @"TestData\gapdh.fasta")]
         public static void TestQuantificationDoesntCrashOnUnsupportedFiles(string filepath, string dbPath)
         {
@@ -54,6 +82,7 @@ public static void TestQuantificationDoesntCrashOnUnsupportedFiles(string filepa
         [Test]
         [TestCase(@"TestData\ok.mgf", @"TestData\okk.xml")]
         [TestCase(@"TestData\snippet.d", @"TestData\gapdh.fasta")]
+        [TestCase(@"TestData\centroid_1x_MS1_4x_autoMS2.d", @"TestData\gapdh.fasta")]
         [TestCase(@"TopDownTestData\JurkatTopDownRep2Fract1_ms2.msalign", @"TestData\gapdh.fasta")]
         public static void TestCalibrationDoesntCrashOnUnsupportedFiles(string filepath, string dbPath)
         {
@@ -88,6 +117,7 @@ public static void TestCalibrationDoesntCrashOnUnsupportedFiles(string filepath,
         [Test]
         [TestCase(@"TestData\ok.mgf", @"TestData\okk.xml")]
         [TestCase(@"TestData\snippet.d", @"TestData\gapdh.fasta")]
+        [TestCase(@"TestData\centroid_1x_MS1_4x_autoMS2.d", @"TestData\gapdh.fasta")]
         [TestCase(@"TopDownTestData\JurkatTopDownRep2Fract1_ms2.msalign", @"TestData\gapdh.fasta")]
         public static void TestAveragingDoesntCrashOnUnsupportedFiles(string filepath, string dbPath)
         {