Merge branch 'develop' into releases/Release_2_11_2_Branch

diff --git a/src/jalview/io/FlatFile.java b/src/jalview/io/EMBLLikeFlatFile.java
similarity index 91%
rename from src/jalview/io/FlatFile.java
rename to src/jalview/io/EMBLLikeFlatFile.java
index 55fdd37..64943b2 100644 (file)
--- a/src/jalview/io/FlatFile.java
+++ b/src/jalview/io/EMBLLikeFlatFile.java
@@ -7,6 +7,7 @@ import java.util.Arrays;
 import java.util.HashMap;
 import java.util.Hashtable;
 import java.util.List;
+import java.util.Locale;
 import java.util.Map;
 import java.util.Map.Entry;
 import java.util.TreeMap;
@@ -44,7 +45,7 @@ import jalview.util.MappingUtils;
  * each line is formatted differently in GenBank and EMBL. See
  * http://www.insdc.org/files/feature_table.html#7.1.
  */
-public abstract class FlatFile extends AlignFile
+public abstract class EMBLLikeFlatFile extends AlignFile
 {
   protected static final String LOCATION = "location";
 
@@ -159,6 +160,13 @@ public abstract class FlatFile extends AlignFile
   }
 
   /*
+   * when true, interpret the mol_type 'source' feature attribute
+   * and generate an RNA sequence from the DNA record
+   */
+  protected boolean produceRna=true;
+    
+
+  /*
    * values parsed from the data file
    */
   protected String sourceDb;
@@ -173,6 +181,8 @@ public abstract class FlatFile extends AlignFile
 
   protected List<DBRefEntry> dbrefs;
 
+  protected boolean sequenceStringIsRNA=false;
+
   protected String sequenceString;
 
   protected Map<String, CdsData> cds;
@@ -184,7 +194,7 @@ public abstract class FlatFile extends AlignFile
    * @param sourceId
    * @throws IOException
    */
-  public FlatFile(FileParse fp, String sourceId) throws IOException
+  public EMBLLikeFlatFile(FileParse fp, String sourceId) throws IOException
   {
     super(false, fp); // don't parse immediately
     this.sourceDb = sourceId;
@@ -199,6 +209,52 @@ public abstract class FlatFile extends AlignFile
   }
 
   /**
+   * process attributes for 'source' until the next FT feature entry
+   * only interested in 'mol_type'
+   * @param tokens
+   * @return
+   * @throws IOException
+   */
+  private String parseSourceQualifiers(String[] tokens) throws IOException
+  {
+    if (!"source".equals(tokens[0]))
+    {
+      throw (new RuntimeException("Not given a 'source' qualifier line"));
+    }
+    // search for mol_type attribute
+
+    StringBuilder sb = new StringBuilder().append(tokens[1]); // extent of
+                                                              // sequence
+
+    String line = parseFeatureQualifier(sb, false);
+    while (line != null)
+    {
+      if (!line.startsWith("FT    ")) // four spaces, end of this feature table
+                                      // entry
+      {
+        return line;
+      }
+
+      // case sensitive ?
+      int p = line.indexOf("\\mol_type");
+      int qs = line.indexOf("\"", p);
+      int qe = line.indexOf("\"", qs + 1);
+      String qualifier=line.substring(qs,qe).toLowerCase(Locale.ROOT);
+      if (qualifier.indexOf("rna") > -1)
+      {
+        sequenceStringIsRNA = true;
+      }
+      if (qualifier.indexOf("dna") > -1)
+      {
+        sequenceStringIsRNA = false;
+      }
+      line=parseFeatureQualifier(sb, false);
+    }
+    return line;
+  }
+
+   
+  /**
    * Parses one (GenBank or EMBL format) CDS feature, saves the parsed data, and
    * returns the next line
    * 
@@ -335,6 +391,12 @@ public abstract class FlatFile extends AlignFile
     {
       name = this.sourceDb + "|" + name;
     }
+
+    if (produceRna && sequenceStringIsRNA)
+    {
+      sequenceString = sequenceString.replace('T', 'U').replace('t', 'u');
+    }
+
     SequenceI seq = new Sequence(name, this.sequenceString);
     seq.setDescription(this.description);
 
@@ -738,11 +800,14 @@ public abstract class FlatFile extends AlignFile
   protected String parseFeature(String line) throws IOException
   {
     String[] tokens = line.trim().split(WHITESPACE);
-    if (tokens.length < 2 || !"CDS".equals(tokens[0]))
+    if (tokens.length < 2 || (!"CDS".equals(tokens[0]) && (!"source".equals(tokens[0]))))
     {
       return nextLine();
     }
-
+    if (tokens[0].equals("source"))
+    {
+      return parseSourceQualifiers(tokens);
+    }
     return parseCDSFeature(tokens[1]);
   }
 }

diff --git a/src/jalview/io/EmblFlatFile.java b/src/jalview/io/EmblFlatFile.java
index 19496ef..7808d1a 100644 (file)
--- a/src/jalview/io/EmblFlatFile.java
+++ b/src/jalview/io/EmblFlatFile.java
@@ -25,7 +25,7 @@ import jalview.util.DBRefUtils;
  * @see ftp://ftp.ebi.ac.uk/pub/databases/ena/sequence/release/doc/usrman.txt
  * @see ftp://ftp.ebi.ac.uk/pub/databases/embl/doc/FT_current.html
  */
-public class EmblFlatFile extends FlatFile
+public class EmblFlatFile extends EMBLLikeFlatFile
 {
   /**
    * Constructor given a data source and the id of the source database

diff --git a/src/jalview/io/GenBankFile.java b/src/jalview/io/GenBankFile.java
index ba7b4b4..f1ca0e3 100644 (file)
--- a/src/jalview/io/GenBankFile.java
+++ b/src/jalview/io/GenBankFile.java
@@ -16,7 +16,7 @@ import java.io.IOException;
  * @author gmcarstairs
  * @see https://www.ncbi.nlm.nih.gov/Sitemap/samplerecord.html
  */
-public class GenBankFile extends FlatFile
+public class GenBankFile extends EMBLLikeFlatFile
 {
   private static final String DEFINITION = "DEFINITION";
 

diff --git a/src/jalview/ws/dbsources/EmblFlatfileSource.java b/src/jalview/ws/dbsources/EmblFlatfileSource.java
index 7d3c6dd..2058800 100644 (file)
--- a/src/jalview/ws/dbsources/EmblFlatfileSource.java
+++ b/src/jalview/ws/dbsources/EmblFlatfileSource.java
@@ -97,7 +97,6 @@ public abstract class EmblFlatfileSource extends EbiFileRetrievedProxy
       file = reply.getAbsolutePath();
       FileParse fp = new FileParse(file, DataSourceType.FILE);
       EmblFlatFile emblParser = new EmblFlatFile(fp, getDbSource());
-      emblParser.parse();
       SequenceI[] seqs = emblParser.getSeqsAsArray();
       if (seqs.length > 0)
       {

diff --git a/test/jalview/io/EmblFlatFileTest.java b/test/jalview/io/EmblFlatFileTest.java
index ee853f3..7775c8f 100644 (file)
--- a/test/jalview/io/EmblFlatFileTest.java
+++ b/test/jalview/io/EmblFlatFileTest.java
@@ -230,6 +230,22 @@ public class EmblFlatFileTest
     }
     assertEquals(uniprotCount, 8);
   }
+  /**
+   * A fairly tough test, using J03321 (circular DNA), which has 8 CDS features,
+   * one of them reverse strand
+   * 
+   * @throws MalformedURLException
+   * @throws IOException
+   */
+  @Test(groups = "Functional")
+  public void testParseToRNA() throws MalformedURLException, IOException
+  {
+    File dataFile = new File("test/jalview/io/J03321_rna.embl.txt");
+    FileParse fp = new FileParse(dataFile, DataSourceType.FILE);
+    EmblFlatFile parser = new EmblFlatFile(fp, "EmblTest");
+    List<SequenceI> seqs = parser.getSeqs();
+    assertTrue(seqs.get(0).getSequenceAsString().indexOf("u")>-1);
+  }
 
   @Test(groups = "Functional")
   public void testParse_codonStartNot1()

diff --git a/test/jalview/io/GenBankFileTest.java b/test/jalview/io/GenBankFileTest.java
index 89f0d0e..97e4754 100644 (file)
--- a/test/jalview/io/GenBankFileTest.java
+++ b/test/jalview/io/GenBankFileTest.java
@@ -42,7 +42,7 @@ public class GenBankFileTest
     File dataFile = new File("test/jalview/io/J03321.gb");
     FileParse fp = new FileParse(dataFile.getAbsolutePath(),
             DataSourceType.FILE);
-    FlatFile parser = new GenBankFile(fp, "GenBankTest");
+    EMBLLikeFlatFile parser = new GenBankFile(fp, "GenBankTest");
     List<SequenceI> seqs = parser.getSeqs();
 
     assertEquals(seqs.size(), 1);