From fe8a6f4ccc5044d3e7653fef1633f50369121368 Mon Sep 17 00:00:00 2001
From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Thu, 27 Aug 2020 15:40:22 +0100
Subject: [PATCH] JAL-3725 exclude stop codon from CDS-to-protein mapping

---
 src/jalview/ws/dbsources/EmblXmlSource.java      |   27 +++++++---------------
 test/jalview/ws/dbsources/EmblXmlSourceTest.java |   23 +++++++++---------
 2 files changed, 20 insertions(+), 30 deletions(-)

diff --git a/src/jalview/ws/dbsources/EmblXmlSource.java b/src/jalview/ws/dbsources/EmblXmlSource.java
index c2d661b..06cbb13 100644
--- a/src/jalview/ws/dbsources/EmblXmlSource.java
+++ b/src/jalview/ws/dbsources/EmblXmlSource.java
@@ -20,8 +20,6 @@
  */
 package jalview.ws.dbsources;
 
-import java.util.Locale;
-
 import java.io.File;
 import java.io.FileInputStream;
 import java.io.InputStream;
@@ -30,6 +28,7 @@ import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.Hashtable;
 import java.util.List;
+import java.util.Locale;
 import java.util.Map;
 import java.util.Map.Entry;
 
@@ -65,12 +64,6 @@ import jalview.xml.binding.embl.EntryType.Feature.Qualifier;
 import jalview.xml.binding.embl.ROOT;
 import jalview.xml.binding.embl.XrefType;
 
-/**
- * Provides XML binding and parsing of EMBL or EMBLCDS records retrieved from
- * (e.g.) {@code https://www.ebi.ac.uk/ena/data/view/x53828&display=xml}.
- * 
- * @deprecated endpoint withdrawn August 2020 (JAL-3692), use EmblFlatfileSource
- */
 public abstract class EmblXmlSource extends EbiFileRetrievedProxy
 {
   private static final Regex ACCESSION_REGEX = new Regex("^[A-Z]+[0-9]+");
@@ -104,8 +97,8 @@ public abstract class EmblXmlSource extends EbiFileRetrievedProxy
     try
     {
       reply = dbFetch.fetchDataAsFile(
-              emprefx.toLowerCase(Locale.ROOT) + ":" + query.trim(), "display=xml",
-              "xml");
+              emprefx.toLowerCase(Locale.ROOT) + ":" + query.trim(),
+              "display=xml", "xml");
     } catch (Exception e)
     {
       stopQuery();
@@ -455,9 +448,8 @@ public abstract class EmblXmlSource extends EbiFileRetrievedProxy
         else
         {
           // final product length truncation check
-          int[] cdsRanges = adjustForProteinLength(translationLength,
-                  exons);
-          dnaToProteinMapping = new Mapping(product, cdsRanges,
+          int[] exons2 = adjustForProteinLength(translationLength, exons);
+          dnaToProteinMapping = new Mapping(product, exons2,
                   new int[]
                   { 1, translationLength }, 3, 1);
           if (product != null)
@@ -759,8 +751,7 @@ public abstract class EmblXmlSource extends EbiFileRetrievedProxy
 
   /**
    * Truncates (if necessary) the exon intervals to match 3 times the length of
-   * the protein; also accepts 3 bases longer (for stop codon not included in
-   * protein)
+   * the protein (including truncation for stop codon included in exon)
    * 
    * @param proteinLength
    * @param exon
@@ -777,11 +768,9 @@ public abstract class EmblXmlSource extends EbiFileRetrievedProxy
     int exonLength = MappingUtils.getLength(Arrays.asList(exon));
 
     /*
-     * if exon length matches protein, or is shorter, or longer by the 
-     * length of a stop codon (3 bases), then leave it unchanged
+     * if exon length matches protein, or is shorter, then leave it unchanged
      */
-    if (expectedCdsLength >= exonLength
-            || expectedCdsLength == exonLength - 3)
+    if (expectedCdsLength >= exonLength)
     {
       return exon;
     }
diff --git a/test/jalview/ws/dbsources/EmblXmlSourceTest.java b/test/jalview/ws/dbsources/EmblXmlSourceTest.java
index a0991e5..236e0a5 100644
--- a/test/jalview/ws/dbsources/EmblXmlSourceTest.java
+++ b/test/jalview/ws/dbsources/EmblXmlSourceTest.java
@@ -26,6 +26,14 @@ import static org.testng.AssertJUnit.assertNull;
 import static org.testng.AssertJUnit.assertSame;
 import static org.testng.AssertJUnit.assertTrue;
 
+import java.io.ByteArrayInputStream;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.List;
+
+import org.testng.annotations.BeforeClass;
+import org.testng.annotations.Test;
+
 import jalview.datamodel.AlignmentI;
 import jalview.datamodel.DBRefEntry;
 import jalview.datamodel.DBRefSource;
@@ -36,14 +44,6 @@ import jalview.xml.binding.embl.EntryType.Feature;
 import jalview.xml.binding.embl.EntryType.Feature.Qualifier;
 import jalview.xml.binding.embl.XrefType;
 
-import java.io.ByteArrayInputStream;
-import java.util.ArrayList;
-import java.util.Arrays;
-import java.util.List;
-
-import org.testng.annotations.BeforeClass;
-import org.testng.annotations.Test;
-
 public class EmblXmlSourceTest
 {
 
@@ -352,11 +352,12 @@ public class EmblXmlSourceTest
     // exact length match:
     assertSame(exons, EmblXmlSource.adjustForProteinLength(6, exons));
 
-    // match if we assume exons include stop codon not in protein:
-    assertSame(exons, EmblXmlSource.adjustForProteinLength(5, exons));
+    // truncate last exon by 3bp (e.g. stop codon)
+    int[] truncated = EmblXmlSource.adjustForProteinLength(5, exons);
+    assertEquals("[11, 15, 21, 25, 31, 35]", Arrays.toString(truncated));
 
     // truncate last exon by 6bp
-    int[] truncated = EmblXmlSource.adjustForProteinLength(4, exons);
+    truncated = EmblXmlSource.adjustForProteinLength(4, exons);
     assertEquals("[11, 15, 21, 25, 31, 32]", Arrays.toString(truncated));
 
     // remove last exon and truncate preceding by 1bp
-- 
1.7.10.2