From: Jim Procter <j.procter@dundee.ac.uk>
Date: Fri, 1 Oct 2021 17:29:57 +0000 (+0100)
Subject: JAL-2909 JAL-3894 hack in support to import regions of an unindexed SAM file.
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;h=refs%2Fheads%2Ffeatures%2FJAL-2909_bamImport_2_11_2;p=jalview.git

JAL-2909 JAL-3894 hack in support to import regions of an unindexed SAM file.
---

diff --git a/src/jalview/datamodel/CigarParser.java b/src/jalview/datamodel/CigarParser.java
index a42c2b1..113223d 100644
--- a/src/jalview/datamodel/CigarParser.java
+++ b/src/jalview/datamodel/CigarParser.java
@@ -9,7 +9,9 @@ import java.util.TreeMap;
 
 import htsjdk.samtools.CigarElement;
 import htsjdk.samtools.CigarOperator;
+import htsjdk.samtools.SAMFormatException;
 import htsjdk.samtools.SAMRecord;
+import jalview.bin.Cache;
 
 public class CigarParser
 {
@@ -291,7 +293,14 @@ public class CigarParser
     do
     {
       // check extent of read
-      SAMRecord rec = it.next();
+      SAMRecord rec = null;
+      try {
+        rec = it.next();
+      } catch (SAMFormatException ex)
+      {
+        Cache.log.info("Bailing on parsing SAM File - see error below",ex);
+        break;
+      }
       if (extent != null)
       {
 
diff --git a/src/jalview/io/BamFile.java b/src/jalview/io/BamFile.java
index 8d55c53..2d079cf 100644
--- a/src/jalview/io/BamFile.java
+++ b/src/jalview/io/BamFile.java
@@ -20,6 +20,7 @@
  */
 package jalview.io;
 
+import jalview.bin.Cache;
 import jalview.datamodel.CigarParser;
 import jalview.datamodel.Range;
 import jalview.datamodel.Sequence;
@@ -35,6 +36,7 @@ import java.util.Map;
 import java.util.PrimitiveIterator.OfInt;
 import java.util.SortedMap;
 
+import htsjdk.samtools.SAMFormatException;
 import htsjdk.samtools.SAMRecord;
 import htsjdk.samtools.SAMRecordIterator;
 import htsjdk.samtools.SAMSequenceRecord;
@@ -60,6 +62,8 @@ public class BamFile extends AlignFile
   // first position in alignment
   private int alignmentStart = -1;
 
+  private File _bamFile;
+
   /**
    * Creates a new BamFile object.
    */
@@ -81,13 +85,28 @@ public class BamFile extends AlignFile
           throws IOException
   {
     super(true, inFile, sourceType);
+    _bamFile = new File(inFile);
+    initFileReader();    
+  }
+  private void initFileReader() throws IOException
+  {
     final SamReaderFactory factory = SamReaderFactory.makeDefault()
             .enable(SamReaderFactory.Option.INCLUDE_SOURCE_IN_RECORDS,
                     SamReaderFactory.Option.VALIDATE_CRC_CHECKSUMS)
             .validationStringency(ValidationStringency.SILENT);
-    fileReader = factory.open(new File(inFile));
+    // File-based bam
+    if (_bamFile!=null)
+    {
+      fileReader = factory.open(_bamFile); // will need to be adapted for JalviewJS/etc
+    }
+    else
+    {
+      // try and locate index ?
+      String index = getDataName() + ".bai";
+      fileReader = factory.open(SamInputResource.of(getDataName())
+              .index(new URL(index)));
+    }
   }
-
   /**
    * Creates a new BamFile object
    * 
@@ -97,25 +116,16 @@ public class BamFile extends AlignFile
    */
   public BamFile(FileParse source) throws IOException
   {
-    super(true, source);
+    super(false, source);
     parseSuffix();
-    final SamReaderFactory factory = SamReaderFactory.makeDefault()
-            .enable(SamReaderFactory.Option.INCLUDE_SOURCE_IN_RECORDS,
-                    SamReaderFactory.Option.VALIDATE_CRC_CHECKSUMS)
-            .validationStringency(ValidationStringency.SILENT);
-
-    // File-based bam
     if (source.getDataSourceType() == DataSourceType.FILE)
-    {
-      fileReader = factory.open(source.inFile);
-    }
-    else
-    {
-      // locate index ?
-      String index = source.getDataName() + ".bai";
-      fileReader = factory.open(SamInputResource.of(source.getDataName())
-              .index(new URL(index)));
-    }
+     {
+       _bamFile = source.inFile;
+     } else {
+       
+     }
+    initFileReader();
+    doParse();
   }
   
   @Override
@@ -166,11 +176,20 @@ public class BamFile extends AlignFile
   @Override
   public void parse()
   {
+    boolean needToReopen=false;
     // only actually parse if params are set
     if (chromosome != null && chromosome != "")
     {
-      SAMRecordIterator it = fileReader.query(chromosome, start, end,
+      SAMRecordIterator it;
+      try {
+        it = fileReader.query(chromosome, start, end,
               false);
+      } catch (UnsupportedOperationException ex)
+      {
+        needToReopen=true;
+        // could be a sam text file, so we just iterate through without query
+        it = fileReader.iterator();
+      }
       CigarParser parser = new CigarParser('-');
       Range[] xtent = new Range[] { new Range(start, end) };
       SortedMap<Integer, Integer> insertions[] = parser.getInsertions(it,
@@ -193,12 +212,35 @@ public class BamFile extends AlignFile
       padRef(refSeq, parser);
       padRef(revRefSeq, parser);
 
-      it = fileReader.query(chromosome, start, end, false);
+      if (needToReopen)
+      {
+        try {
+          initFileReader();
+        } catch (IOException x)
+        {
+          Cache.log.warn("Couldn't reopen S/BAM file",x);
+        }
+      }
+      try {
+        it = fileReader.query(chromosome, start, end,
+              false);
+      } catch (UnsupportedOperationException ex)
+      {
+        // could be a sam text file, so we just iterate through without query
+        it = fileReader.iterator();
+      }
 
       ArrayList<SequenceI> fwd = new ArrayList(), rev = new ArrayList();
       while (it.hasNext())
       {
-        SAMRecord rec = it.next();
+        SAMRecord rec=null;
+        try {
+          rec = it.next();
+        } catch (SAMFormatException q)
+        {
+          Cache.log.info("Bailing on bad SAM line again",q);
+          break;
+        }
 
         // set the alignment start to be start of first read (we assume reads
         // are sorted)
diff --git a/src/jalview/io/FileFormat.java b/src/jalview/io/FileFormat.java
index 1731eaf..66bcbde 100644
--- a/src/jalview/io/FileFormat.java
+++ b/src/jalview/io/FileFormat.java
@@ -379,7 +379,7 @@ public enum FileFormat implements FileFormatI
       return true;
     }
   },
-  Bam("bam", "bam", true, false)
+  Bam("bam", "bam, sam", true, false)
   {
     @Override
     public AlignmentFileReaderI getReader(FileParse source)
diff --git a/src/jalview/io/IdentifyFile.java b/src/jalview/io/IdentifyFile.java
index 0c201e1..55f5651 100755
--- a/src/jalview/io/IdentifyFile.java
+++ b/src/jalview/io/IdentifyFile.java
@@ -207,6 +207,11 @@ public class IdentifyFile
             break;
           }
         }
+        if (data.startsWith("@HD\t"))
+        {
+          reply = FileFormat.Bam;
+          break;
+        }
         if (data.startsWith("H ") && !aaIndexHeaderRead)
         {
           aaIndexHeaderRead = true;