MS Levels

new key -L / --msLevel allowing to select which spectra levels to output

Author: caetera

MainClass.cs

@@ -9,6 +9,8 @@
 using ThermoRawFileParser.Query;
 using ThermoRawFileParser.XIC;
 using System.Globalization;
+using System.Collections.Generic;
+using System.Text.RegularExpressions;
 
 namespace ThermoRawFileParser
 {
@@ -422,6 +424,11 @@ private static void RegularParametersParsing(string[] args)
                     "e|ignoreInstrumentErrors", "Ignore missing properties by the instrument.",
                     v => parseInput.IgnoreInstrumentErrors = v != null
                 },
+                {
+                    "L=|msLevel=",
+                    "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)
+                },
                 {
                     "u:|s3_url:",
                     "Optional property to write directly the data into S3 Storage.",
@@ -756,5 +763,64 @@ private static void ShowHelp(string message, OptionException optionException, Op
             optionSet.WriteOptionDescriptions(Console.Error);
             Environment.Exit(-1);
         }
+
+        private static HashSet<int> ParseMsLevel(string inputString)
+        {
+            HashSet<int> result = new HashSet<int>();
+            Regex valid = new Regex(@"^[\d,\-\s]+$");
+            Regex interval = new Regex(@"^\s*(\d+)?\s*(-)?\s*(\d+)?\s*$");
+
+            if (!valid.IsMatch(inputString))
+                throw new OptionException("Invalid characters in msLevel key", "msLevel");
+
+            foreach (var piece in inputString.Split(new char[] { ',' }))
+            {
+                try
+                {
+                    int start;
+                    int end;
+
+                    var intervalMatch = interval.Match(piece);
+
+                    if (!intervalMatch.Success)
+                        throw new OptionException();
+
+                    if (intervalMatch.Groups[2].Success) //it is interval
+                    {
+                        if (intervalMatch.Groups[1].Success)
+                            start = Math.Max(1, int.Parse(intervalMatch.Groups[1].Value));
+                        else
+                            start = 1;
+
+                        if (intervalMatch.Groups[3].Success)
+                            end = Math.Min(10, int.Parse(intervalMatch.Groups[3].Value));
+                        else
+                            end = 10;
+                    }
+                    else
+                    {
+                        if (intervalMatch.Groups[1].Success)
+                            end = start = int.Parse(intervalMatch.Groups[1].Value);
+                        else
+                            throw new OptionException();
+
+                        if (intervalMatch.Groups[3].Success)
+                            throw new OptionException();
+                    }
+
+                    for (int l = start; l <= end; l++)
+                    {
+                        result.Add(l);
+                    }
+                }
+
+                catch (Exception ex)
+                {
+                    throw new OptionException(String.Format("Cannot parse part of msLevel input: '{0}'", piece), "msLevel", ex);
+                }
+            }
+
+            return result;
+        }
     }
 }

ParseInput.cs

@@ -1,11 +1,15 @@
 using System;
+using System.Collections.Generic;
 using System.IO;
 using ThermoRawFileParser.Writer;
 
 namespace ThermoRawFileParser
 {
     public class ParseInput
     {
+        //all ms levels
+        private readonly HashSet<int> allLevels = new HashSet<int>(new int[] { 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 });
+
         /// <summary>
         /// The RAW file path.
         /// </summary>
@@ -71,6 +75,8 @@ public string RawFilePath
 
         public bool IgnoreInstrumentErrors { get; set; }
 
+        public HashSet<int> MsLevel { get; set; }
+
         private S3Loader S3Loader { get; set; }
 
         public string S3AccessKeyId { get; set; }
@@ -101,6 +107,7 @@ public ParseInput()
             LogFormat = LogFormat.DEFAULT;
             IgnoreInstrumentErrors = false;
             AllDetectors = false;
+            MsLevel = allLevels;
         }
 
         public ParseInput(string rawFilePath, string rawDirectoryPath, string outputDirectory, OutputFormat outputFormat

Writer/MgfSpectrumWriter.cs

@@ -20,6 +20,7 @@ public class MgfSpectrumWriter : SpectrumWriter
 
         public MgfSpectrumWriter(ParseInput parseInput) : base(parseInput)
         {
+            ParseInput.MsLevel.Remove(1); //MS1 spectra are not supposed to be in MGF
         }
 
         /// <inheritdoc />       
@@ -60,7 +61,7 @@ public override void Write(IRawDataPlus rawFile, int firstScanNumber, int lastSc
                     var scanEvent = rawFile.GetScanEventForScanNumber(scanNumber);
 
                     // don't include MS1 spectra
-                    if (scanFilter.MSOrder != MSOrderType.Ms)
+                    if (ParseInput.MsLevel.Contains((int)scanFilter.MSOrder))
                     {
                         IReaction reaction = GetReaction(scanEvent, scanNumber);
 

Writer/MzMlSpectrumWriter.cs

@@ -334,7 +334,8 @@ public override void Write(IRawDataPlus rawFile, int firstScanNumber, int lastSc
                     }
 
                     var spectrum = ConstructMSSpectrum(scanNumber);
-                    if (spectrum != null)
+                    var level = int.Parse(spectrum.cvParam.Where(p => p.accession == "MS:1000511").First().value);
+                    if (spectrum != null && ParseInput.MsLevel.Contains(level))
                     {
                         spectrum.index = index.ToString();
                         if (_doIndexing)