From: Jim Procter <jprocter@issues.jalview.org>
Date: Wed, 17 Feb 2021 17:16:03 +0000 (+0000)
Subject: JAL-3821 patch for 2.11.2 to retrieve RNA ENA records as RNA
X-Git-Tag: Release_2_11_2_0~34^2~33
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;h=36952139fb50d6ee1d12ecf69f340986ae9bafcc;p=jalview.git

JAL-3821 patch for 2.11.2 to retrieve RNA ENA records as RNA
---

diff --git a/src/jalview/io/EmblFlatFile.java b/src/jalview/io/EmblFlatFile.java
index 900aef8..ff18a34 100644
--- a/src/jalview/io/EmblFlatFile.java
+++ b/src/jalview/io/EmblFlatFile.java
@@ -50,6 +50,11 @@ public class EmblFlatFile extends AlignFile // FileParse
   private static final String DOUBLED_QUOTE = QUOTE + QUOTE;
 
   /**
+   * when true, interpret the mol_type 'source' feature attribute
+   * and generate an RNA sequence from the DNA record
+   */
+  private boolean produceRna=true;
+  /**
    * A data bean class to hold values parsed from one CDS Feature (FT)
    */
   class CdsData
@@ -86,6 +91,7 @@ public class EmblFlatFile extends AlignFile // FileParse
 
   private List<DBRefEntry> dbrefs; // from DR
 
+  private boolean sequenceStringIsRNA=false;
   private String sequenceString; // from SQ lines
 
   /*
@@ -317,11 +323,19 @@ public class EmblFlatFile extends AlignFile // FileParse
   String parseFT(String line) throws IOException
   {
     String[] tokens = line.split(WHITESPACE);
-    if (tokens.length < 3 || !"CDS".equals(tokens[1]))
+    if (tokens.length < 3 || (!"CDS".equals(tokens[1]) && !"source".equals(tokens[1])))
     {
       return nextLine();
     }
+    
+    if (tokens[1].equals("source"))
+    {
+      return parseSourceQualifiers(tokens);
+    }
 
+    /*
+     * parse location - which may be over more than one line e.g. EAW51554
+     */
     CdsData data = new CdsData();
     data.cdsLocation = tokens[2];
     // TODO location can be over >1 line e.g. EAW51554
@@ -418,6 +432,50 @@ public class EmblFlatFile extends AlignFile // FileParse
   }
 
   /**
+   * 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[1]))
+    {
+      throw (new RuntimeException("Not given a source qualifier"));
+    }
+    // search for mol_type attribute
+
+    StringBuilder sb = new StringBuilder().append(tokens[2]); // extent of
+                                                              // sequence
+
+    String line = parseFeatureQualifier(sb, "source");
+    while (line != null)
+    {
+      if (!line.startsWith("FT    ")) // four spaces, end of this feature table
+                                      // entry
+      {
+        return line;
+      }
+
+      int p = line.indexOf("\\mol_type");
+      int qs = line.indexOf("\"", p);
+      int qe = line.indexOf("\"", qs + 1);
+      String qualifier=line.substring(qs,qe).toLowerCase();
+      if (qualifier.indexOf("rna") > -1)
+      {
+        sequenceStringIsRNA = true;
+      }
+      if (qualifier.indexOf("dna") > -1)
+      {
+        sequenceStringIsRNA = false;
+      }
+      line=parseFeatureQualifier(sb, "source");
+    }
+    return line;
+  }
+
+  /**
    * Removes leading or trailing double quotes (") unless doubled, and changes
    * any 'escaped' (doubled) double quotes to single characters. As per the
    * Feature Table specification for Qualifiers, Free Text.
@@ -524,6 +582,12 @@ public class EmblFlatFile extends AlignFile // FileParse
     {
       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);