perf: replace jmzml JAXB parser with StAX-based mzML reader

pom.xml

@@ -38,8 +38,6 @@
                     <archive>
                         <manifest>
                             <addClasspath>true</addClasspath>
-                        </manifest>
-                        <manifest>
                             <mainClass>edu.ucsd.msjava.ui.MSGFPlus</mainClass>
                         </manifest>
                     </archive>
@@ -79,7 +77,6 @@
                             <goal>shade</goal>
                         </goals>
                         <configuration>
-                            <!-- <finalName>${project.artifactId}-${project.version}-jar-w-reduced-deps</finalName> -->
                             <finalName>${project.artifactId}</finalName>
                             <transformers>
                                 <transformer implementation="org.apache.maven.plugins.shade.resource.ManifestResourceTransformer">
@@ -95,14 +92,10 @@
                                         <exclude>META-INF/*.RSA</exclude>
                                     </excludes>
                                 </filter>
-                                <!-- fastutil shade filter removed: jmzml 1.7.11 requires
-                                     additional fastutil classes beyond the original whitelist.
-                                     Full fastutil 8.5.12 is now included as an explicit dependency. -->
                                 <filter>
                                     <artifact>*:*</artifact>
                                     <excludes>
                                         <exclude>**/.svn/**</exclude>
-                                        <!-- <exclude>**/*.java</exclude> -->
                                     </excludes>
                                 </filter>
                             </filters>
@@ -124,34 +117,6 @@
                 </exclusion>
             </exclusions>
         </dependency>
-        <dependency>
-            <groupId>uk.ac.ebi.pride.tools</groupId>
-            <artifactId>jmzreader</artifactId>
-            <version>2.0.6</version>
-            <type>jar</type>
-            <exclusions>
-                <exclusion>
-                    <groupId>com.sun.xml.bind</groupId>
-                    <artifactId>jaxb-xjc</artifactId>
-                </exclusion>
-                <exclusion>
-                    <groupId>javax.xml.bind</groupId>
-                    <artifactId>jaxb-api</artifactId>
-                </exclusion>
-                <exclusion>
-                    <groupId>com.sun.xml.bind</groupId>
-                    <artifactId>jaxb-core</artifactId>
-                </exclusion>
-                <exclusion>
-                    <groupId>com.sun.xml.bind</groupId>
-                    <artifactId>jaxb-impl</artifactId>
-                </exclusion>
-                <exclusion>
-                    <groupId>com.sun.activation</groupId>
-                    <artifactId>javax.activation</artifactId>
-                </exclusion>
-            </exclusions>
-        </dependency>
         <dependency>
             <groupId>junit</groupId>
             <artifactId>junit</artifactId>
@@ -175,11 +140,8 @@
             <artifactId>commons-text</artifactId>
             <version>1.11.0</version>
         </dependency>
-        <dependency>
-            <groupId>uk.ac.ebi.jmzml</groupId>
-            <artifactId>jmzml</artifactId>
-            <version>1.7.11</version>
-        </dependency>
+        <!-- jmzml removed: replaced by StaxMzMLParser -->
+        <!-- fastutil is required at runtime by jmzidentml -->
         <dependency>
             <groupId>it.unimi.dsi</groupId>
             <artifactId>fastutil</artifactId>

src/main/java/edu/ucsd/msjava/msdbsearch/SearchParams.java

@@ -53,6 +53,7 @@ public class SearchParams {
     private boolean allowDenseCentroidedPeaks;
     private int minMSLevel;
     private int maxMSLevel;
+    private int outputFormat; // 0=mzid, 1=tsv, 2=both
 
     public SearchParams() {
     }
@@ -232,6 +233,19 @@ public int getMaxMSLevel() {
         return maxMSLevel;
     }
 
+    /** 0=mzid, 1=tsv, 2=both */
+    public int getOutputFormat() {
+        return outputFormat;
+    }
+
+    public boolean writeMzid() {
+        return outputFormat == 0 || outputFormat == 2;
+    }
+
+    public boolean writeTsv() {
+        return outputFormat == 1 || outputFormat == 2;
+    }
+
     /**
      * Look for # in dataLine
      * If present, remove that character and any comment after it
@@ -421,6 +435,7 @@ public String parse(ParamManager paramManager) {
         }
 
         allowDenseCentroidedPeaks = paramManager.getAllowDenseCentroidedPeaks() == 1;
+        outputFormat = paramManager.getOutputFormat();
 
         IntRangeParameter msLevelParam = paramManager.getMSLevelParameter();
         minMSLevel = msLevelParam.getMin();

src/main/java/edu/ucsd/msjava/msutil/SpecFileFormat.java

@@ -22,7 +22,6 @@ public String getPSIName() {
     }
 
     public static final SpecFileFormat MGF;
-    public static final SpecFileFormat MZXML;
     public static final SpecFileFormat MZML;
     public static final SpecFileFormat MS2;
     public static final SpecFileFormat PKL;
@@ -44,7 +43,6 @@ public static SpecFileFormat getSpecFileFormat(String specFileName) {
 
     static {
         MGF = new SpecFileFormat(".mgf", "MS:1001062", "Mascot MGF file");
-        MZXML = new SpecFileFormat(".mzXML", "MS:1000566", "ISB mzXML file");
         MZML = new SpecFileFormat(".mzML", "MS:1000584", "mzML file");
         MS2 = new SpecFileFormat(".ms2", "MS:1001466", "MS2 file");
         PKL = new SpecFileFormat(".pkl", "MS:1000565", "Micromass PKL file");
@@ -53,7 +51,6 @@ public static SpecFileFormat getSpecFileFormat(String specFileName) {
 
         specFileFormatList = new ArrayList<SpecFileFormat>();
         specFileFormatList.add(MGF);
-        specFileFormatList.add(MZXML);
         specFileFormatList.add(MZML);
         specFileFormatList.add(MS2);
         specFileFormatList.add(PKL);

src/main/java/edu/ucsd/msjava/msutil/SpecKey.java

@@ -1,6 +1,5 @@
 package edu.ucsd.msjava.msutil;
 
-import edu.ucsd.msjava.parser.MzXMLSpectraIterator;
 import edu.ucsd.msjava.parser.SpectrumParser;
 
 import java.util.ArrayList;
@@ -292,20 +291,5 @@ public static ArrayList<SpecKey> getFusedSpecKeyList(Iterator<Spectrum> itr, int
         return specKeyList;
     }
 
-    public static void main(String argv[]) throws Exception {
-        test();
-    }
-
-    public static void test() throws Exception {
-        String fileName = "/home/sangtaekim/Research/Data/HeckRevision/CIDETDPairs/mzXML/090121_NM_Trypsin_20.mzXML";
-        int minCharge = 2, maxCharge = 3;
-        MzXMLSpectraIterator itr = new MzXMLSpectraIterator(fileName);
-        ArrayList<SpecKey> list = SpecKey.getFusedSpecKeyList(itr, 0, Integer.MAX_VALUE, minCharge, maxCharge);
-        for (SpecKey specKey : list) {
-            if (specKey.getSpecIndexList() == null || specKey.getSpecIndexList().size() != 2)
-                System.out.println(specKey.getSpecKeyString() + "\t" + specKey.getSpecIndexList());
-        }
-        System.out.println("Size: " + list.size());
-    }
 
 }

src/main/java/edu/ucsd/msjava/msutil/SpectraAccessor.java

@@ -1,9 +1,9 @@
 package edu.ucsd.msjava.msutil;
 
 import edu.ucsd.msjava.mzid.Constants;
-import edu.ucsd.msjava.mzml.MzMLAdapter;
-import edu.ucsd.msjava.mzml.MzMLSpectraIterator;
-import edu.ucsd.msjava.mzml.MzMLSpectraMap;
+import edu.ucsd.msjava.mzml.StaxMzMLParser;
+import edu.ucsd.msjava.mzml.StaxMzMLSpectraIterator;
+import edu.ucsd.msjava.mzml.StaxMzMLSpectraMap;
 import edu.ucsd.msjava.parser.*;
 import uk.ac.ebi.jmzidml.model.mzidml.CvParam;
 
@@ -18,7 +18,7 @@ public class SpectraAccessor {
 
     private SpectrumParser spectrumParser;
 
-    private MzMLAdapter mzmlAdapter = null;
+    private StaxMzMLParser staxParser = null;
 
     private int minMSLevel = 2;
     private int maxMSLevel = 2;
@@ -43,6 +43,9 @@ public SpectraAccessor(File specFile) {
      * @param specFormat
      */
     public SpectraAccessor(File specFile, SpecFileFormat specFormat) {
+        if (specFormat == null) {
+            throw new IllegalArgumentException("Unsupported spectrum file format: " + specFile.getName());
+        }
         this.specFile = specFile;
         this.specFormat = specFormat;
         this.spectrumParser = null;
@@ -61,13 +64,15 @@ public void setMSLevelRange(int minMSLevel, int maxMSLevel) {
 
     public SpectrumAccessorBySpecIndex getSpecMap() {
         if (specMap == null) {
-            if (specFormat == SpecFileFormat.MZXML)
-                specMap = new MzXMLSpectraMap(specFile.getPath()).msLevel(minMSLevel, maxMSLevel);
-            else if (specFormat == SpecFileFormat.MZML) {
-                if (mzmlAdapter == null)
-                    mzmlAdapter = new MzMLAdapter(specFile);
-                mzmlAdapter.msLevel(minMSLevel, maxMSLevel);
-                specMap = new MzMLSpectraMap(mzmlAdapter);
+            if (specFormat == SpecFileFormat.MZML) {
+                if (staxParser == null) {
+                    try {
+                        staxParser = new StaxMzMLParser(specFile);
+                    } catch (Exception e) {
+                        throw new RuntimeException("Failed to parse mzML file: " + specFile.getAbsolutePath(), e);
+                    }
+                }
+                specMap = new StaxMzMLSpectraMap(staxParser, minMSLevel, maxMSLevel);
             } else if (specFormat == SpecFileFormat.DTA_TXT)
                 specMap = new PNNLSpectraMap(specFile.getPath());
             else {
@@ -96,13 +101,15 @@ else if (specFormat == SpecFileFormat.PKL)
 
     public Iterator<Spectrum> getSpecItr() {
         if (specItr == null) {
-            if (specFormat == SpecFileFormat.MZXML)
-                specItr = new MzXMLSpectraIterator(specFile.getPath(), minMSLevel, maxMSLevel);
-            else if (specFormat == SpecFileFormat.MZML) {
-                if (mzmlAdapter == null)
-                    mzmlAdapter = new MzMLAdapter(specFile);
-                mzmlAdapter.msLevel(minMSLevel, maxMSLevel);
-                specItr = new MzMLSpectraIterator(mzmlAdapter);
+            if (specFormat == SpecFileFormat.MZML) {
+                if (staxParser == null) {
+                    try {
+                        staxParser = new StaxMzMLParser(specFile);
+                    } catch (Exception e) {
+                        throw new RuntimeException("Failed to parse mzML file: " + specFile.getAbsolutePath(), e);
+                    }
+                }
+                specItr = new StaxMzMLSpectraIterator(staxParser, minMSLevel, maxMSLevel);
             } else if (specFormat == SpecFileFormat.DTA_TXT)
                 try {
                     specItr = new PNNLSpectraIterator(specFile.getPath());
@@ -168,14 +175,23 @@ public CvParam getSpectrumIDFormatCvParam() {
                 || specFormat == SpecFileFormat.MS2
         )
             cvParam = Constants.makeCvParam("MS:1000774", "multiple peak list nativeID format");
-        else if (specFormat == SpecFileFormat.MZXML)
-            cvParam = Constants.makeCvParam("MS:1000776", "scan number only nativeID format");
         else if (specFormat == SpecFileFormat.MZDATA)
             cvParam = Constants.makeCvParam("MS:1000777", "spectrum identifier nativeID format");
         else if (specFormat == SpecFileFormat.MZML) {
-            if (mzmlAdapter == null)
-                mzmlAdapter = new MzMLAdapter(specFile);
-            cvParam = mzmlAdapter.getSpectrumIDFormatCvParam();
+            if (staxParser == null) {
+                try {
+                    staxParser = new StaxMzMLParser(specFile);
+                } catch (Exception e) {
+                    throw new RuntimeException("Failed to parse mzML file: " + specFile.getAbsolutePath(), e);
+                }
+            }
+            String[] idFormat = staxParser.detectSpectrumIDFormat();
+            if (idFormat != null) {
+                cvParam = Constants.makeCvParam(idFormat[0], idFormat[1]);
+            } else {
+                throw new IllegalStateException("Unsupported mzML format: " + specFile.getAbsolutePath()
+                        + " does not contain a child term of MS:1000767 (native spectrum identifier format)");
+            }
         }
 
         return cvParam;

src/main/java/edu/ucsd/msjava/msutil/Spectrum.java

@@ -1025,8 +1025,8 @@ public static SpecFileFormat getSpectrumFileFormat(String specFileName) {
         int posDot = specFileName.lastIndexOf('.');
         if (posDot >= 0) {
             String extension = specFileName.substring(posDot);
-            if (extension.equalsIgnoreCase(".mzXML") || extension.equalsIgnoreCase(".mzML"))
-                specFormat = SpecFileFormat.MZXML;
+            if (extension.equalsIgnoreCase(".mzML"))
+                specFormat = SpecFileFormat.MZML;
             else if (extension.equalsIgnoreCase(".mgf"))
                 specFormat = SpecFileFormat.MGF;
             else if (extension.equalsIgnoreCase(".ms2"))