Precursor Title in MGF

new flag -P|--mgfPrecursor

Author: caetera

MainClass.cs

@@ -429,6 +429,11 @@ private static void RegularParametersParsing(string[] args)
                     "MS Levels (i.e. MS1, MS2 etc) to include in the output, should be comma separated list of integers ( 1,2,3 ) or intervals ( 1-3 ), open end intervals ( 1- ) are allowed",
                     v => parseInput.MsLevel = ParseMsLevel(v)
                 },
+                {
+                    "P|mgfPrecursor",
+                    "Write precursor scan in MGF file",
+                    v => parseInput.MGFPrecursor = v != null
+                },
                 {
                     "u:|s3_url:",
                     "Optional property to write directly the data into S3 Storage.",

ParseInput.cs

@@ -77,6 +77,8 @@ public string RawFilePath
 
         public HashSet<int> MsLevel { get; set; }
 
+        public bool MGFPrecursor { get; set; }
+
         private S3Loader S3Loader { get; set; }
 
         public string S3AccessKeyId { get; set; }

Writer/MgfSpectrumWriter.cs

@@ -1,10 +1,14 @@
 using System;
 using System.Globalization;
+using System.Linq;
 using System.Reflection;
+using System.Text.RegularExpressions;
 using log4net;
+using ThermoFisher.CommonCore.Data;
 using ThermoFisher.CommonCore.Data.Business;
 using ThermoFisher.CommonCore.Data.FilterEnums;
 using ThermoFisher.CommonCore.Data.Interfaces;
+using ThermoRawFileParser.Util;
 
 namespace ThermoRawFileParser.Writer
 {
@@ -16,6 +20,15 @@ public class MgfSpectrumWriter : SpectrumWriter
         private const string PositivePolarity = "+";
         private const string NegativePolarity = "-";
 
+        //filter string
+        private const string FilterStringIsolationMzPattern = @"ms2 (.*?)@";
+
+        //precursor scan number for MS2 scans
+        private int _precursorMs1ScanNumber;
+
+        // Precursor scan number (value) and isolation m/z (key) for reference in the precursor element of an MS3 spectrum
+        private readonly LimitedSizeDictionary<string, int> _precursorMs2ScanNumbers = new LimitedSizeDictionary<string, int>(40);
+
         // Precursor scan number for reference in the precursor element of an MS2 spectrum
 
         public MgfSpectrumWriter(ParseInput parseInput) : base(parseInput)
@@ -36,7 +49,7 @@ public override void Write(IRawDataPlus rawFile, int firstScanNumber, int lastSc
                 {
                     if (ParseInput.LogFormat == LogFormat.DEFAULT)
                     {
-                        var scanProgress = (int) ((double) scanNumber / (lastScanNumber - firstScanNumber + 1) * 100);
+                        var scanProgress = (int)((double)scanNumber / (lastScanNumber - firstScanNumber + 1) * 100);
                         if (scanProgress % ProgressPercentageStep == 0)
                         {
                             if (scanProgress != lastScanProgress)
@@ -60,6 +73,52 @@ public override void Write(IRawDataPlus rawFile, int firstScanNumber, int lastSc
                     // Get the scan event for this scan number
                     var scanEvent = rawFile.GetScanEventForScanNumber(scanNumber);
 
+                    // precursor reference
+                    var spectrumRef = "";
+
+                    //keeping track of precursor scan
+                    switch (scanFilter.MSOrder)
+                    {
+                        case MSOrderType.Ms:
+
+                            // Keep track of scan number for precursor reference
+                            _precursorMs1ScanNumber = scanNumber;
+
+                            break;
+                        case MSOrderType.Ms2:
+                            // Keep track of scan number and isolation m/z for precursor reference                   
+                            var result = Regex.Match(scanEvent.ToString(), FilterStringIsolationMzPattern);
+                            if (result.Success)
+                            {
+                                if (_precursorMs2ScanNumbers.ContainsKey(result.Groups[1].Value))
+                                {
+                                    _precursorMs2ScanNumbers.Remove(result.Groups[1].Value);
+                                }
+
+                                _precursorMs2ScanNumbers.Add(result.Groups[1].Value, scanNumber);
+                            }
+
+                            spectrumRef = ConstructSpectrumTitle((int)Device.MS, 1, _precursorMs1ScanNumber);
+                            break;
+
+                        case MSOrderType.Ms3:
+                            var precursorMs2ScanNumber = _precursorMs2ScanNumbers.Keys.FirstOrDefault(
+                                isolationMz => scanEvent.ToString().Contains(isolationMz));
+                            if (!precursorMs2ScanNumber.IsNullOrEmpty())
+                            {
+                                spectrumRef = ConstructSpectrumTitle((int)Device.MS, 1, _precursorMs2ScanNumbers[precursorMs2ScanNumber]);
+                            }
+                            else
+                            {
+                                throw new InvalidOperationException("Couldn't find a MS2 precursor scan for MS3 scan " + scanEvent);
+                            }
+                            break;
+
+                        default:
+                            break;
+                    }
+
+
                     // don't include MS1 spectra
                     if (ParseInput.MsLevel.Contains((int)scanFilter.MSOrder))
                     {
@@ -70,7 +129,6 @@ public override void Write(IRawDataPlus rawFile, int firstScanNumber, int lastSc
                         Writer.WriteLine($"SCANS={scanNumber}");
                         Writer.WriteLine(
                             $"RTINSECONDS={(time * 60).ToString(CultureInfo.InvariantCulture)}");
-
                         // trailer extra data list
                         var trailerData = rawFile.GetTrailerExtraInformation(scanNumber);
                         int? charge = null;
@@ -92,7 +150,7 @@ public override void Write(IRawDataPlus rawFile, int firstScanNumber, int lastSc
                                     CultureInfo.CurrentCulture);
                             }
 
-                            if (trailerData.Labels[i] == "MS" + (int) scanFilter.MSOrder + " Isolation Width:")
+                            if (trailerData.Labels[i] == "MS" + (int)scanFilter.MSOrder + " Isolation Width:")
                             {
                                 isolationWidth = double.Parse(trailerData.Values[i], NumberStyles.Any,
                                     CultureInfo.CurrentCulture);
@@ -121,6 +179,8 @@ public override void Write(IRawDataPlus rawFile, int firstScanNumber, int lastSc
                             Writer.WriteLine($"CHARGE={charge}{polarity}");
                         }
 
+                        if(ParseInput.MGFPrecursor) Writer.WriteLine($"PRECURSORTITLE={spectrumRef}");
+
                         // write the filter string
                         //Writer.WriteLine($"SCANEVENT={scanEvent.ToString()}");
 

Writer/MzMlSpectrumWriter.cs

@@ -334,8 +334,10 @@ public override void Write(IRawDataPlus rawFile, int firstScanNumber, int lastSc
                     }
 
                     var spectrum = ConstructMSSpectrum(scanNumber);
+                    
                     var level = int.Parse(spectrum.cvParam.Where(p => p.accession == "MS:1000511").First().value);
-                    if (spectrum != null && ParseInput.MsLevel.Contains(level))
+                    
+                    if (spectrum != null && ParseInput.MsLevel.Contains(level)) //applying MS level filter
                     {
                         spectrum.index = index.ToString();
                         if (_doIndexing)