From b9cc00124dff0ad97f8b154f16d14ef530d03120 Mon Sep 17 00:00:00 2001
From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Thu, 21 Sep 2017 14:13:28 +0100
Subject: [PATCH] JAL-2738 GeneLoci holds mapping of sequence to chromosome

---
 src/jalview/datamodel/GeneLoci.java |   80 +++++++++++++++++++++++++++++++++++
 src/jalview/datamodel/Sequence.java |   57 +++++++++++++++++++++++--
 2 files changed, 133 insertions(+), 4 deletions(-)
 create mode 100644 src/jalview/datamodel/GeneLoci.java

diff --git a/src/jalview/datamodel/GeneLoci.java b/src/jalview/datamodel/GeneLoci.java
new file mode 100644
index 0000000..9f3520f
--- /dev/null
+++ b/src/jalview/datamodel/GeneLoci.java
@@ -0,0 +1,80 @@
+package jalview.datamodel;
+
+import jalview.util.MapList;
+
+/**
+ * A data bean to model one or more contiguous regions on one chromosome
+ */
+public class GeneLoci
+{
+  /*
+   * implemented as an adapter over DBRefEntry with
+   * source -> species id
+   * version -> reference
+   * accession -> chromosome
+   */
+  private DBRefEntry loci;
+
+  boolean forwardStrand;
+
+  /**
+   * Constructor
+   * 
+   * @param taxon
+   * @param ref
+   * @param chrId
+   * @param map
+   * @param forward
+   */
+  public GeneLoci(String taxon, String ref, String chrId, MapList map,
+          boolean forward)
+  {
+    loci = new DBRefEntry(taxon, ref, chrId, new Mapping(map));
+    forwardStrand = forward;
+  }
+
+  /**
+   * Answers the identifier for the species
+   * 
+   * @return
+   */
+  public String getSpecies()
+  {
+    return loci.getSource();
+  }
+
+  /**
+   * Answers the identifier for the genomic reference assembly
+   */
+  public String getReference()
+  {
+    return loci.getVersion();
+  }
+
+  /**
+   * Answers the chromosome identifier
+   * 
+   * @return
+   */
+  public String getChromosome()
+  {
+    return loci.getAccessionId();
+  }
+
+  /**
+   * Answers the mapping from sequence positions (in sequence start..end
+   * coordinates) to the corresponding loci in the chromosome (in reference
+   * assembly coordinates, base 1)
+   * 
+   * @return
+   */
+  public MapList getMapping()
+  {
+    return loci.getMap().getMap();
+  }
+
+  public boolean isForwardStrand()
+  {
+    return forwardStrand;
+  }
+}
diff --git a/src/jalview/datamodel/Sequence.java b/src/jalview/datamodel/Sequence.java
index 2f1da7f..cf1cf94 100755
--- a/src/jalview/datamodel/Sequence.java
+++ b/src/jalview/datamodel/Sequence.java
@@ -106,6 +106,8 @@ public class Sequence extends ASequence implements SequenceI
    */
   private int changeCount;
 
+  private GeneLoci geneLoci;
+
   /**
    * Creates a new Sequence object.
    * 
@@ -645,21 +647,68 @@ public class Sequence extends ASequence implements SequenceI
   }
 
   /**
-   * DOCUMENT ME!
+   * Sets the sequence description, and also parses out any special formats of
+   * interest
    * 
    * @param desc
-   *          DOCUMENT ME!
    */
   @Override
   public void setDescription(String desc)
   {
     this.description = desc;
+    parseDescription();
   }
 
   /**
-   * DOCUMENT ME!
+   * Parses and saves fields of an Ensembl-style description e.g.
+   * chromosome:GRCh38:17:45051610:45109016:1
+   */
+  protected void parseDescription()
+  {
+    if (description == null)
+    {
+      return;
+    }
+    String[] tokens = description.split(":");
+    if (tokens.length == 6 && "chromosome".equals(tokens[0])) {
+      String ref = tokens[1];
+      String chrom = tokens[2];
+      try {
+        int chStart = Integer.parseInt(tokens[3]);
+        int chEnd = Integer.parseInt(tokens[4]);
+        boolean forwardStrand = "1".equals(tokens[5]);
+        String species = ""; // dunno yet!
+        int[] from = new int[] { start, end };
+        int[] to = new int[] { chStart, chEnd };
+        MapList map = new MapList(from, to, 1, 1);
+        GeneLoci gl = new GeneLoci(species, ref, chrom, map, forwardStrand);
+        setGeneLoci(gl);
+      } catch (NumberFormatException e)
+      {
+        System.err.println("Bad integers in description " + description);
+      }
+    }
+  }
+
+  public void setGeneLoci(GeneLoci gl)
+  {
+    geneLoci = gl;
+  }
+
+  /**
+   * Returns the gene loci mapping for the sequence (may be null)
    * 
-   * @return DOCUMENT ME!
+   * @return
+   */
+  public GeneLoci getGeneLoci()
+  {
+    return geneLoci;
+  }
+
+  /**
+   * Answers the description
+   * 
+   * @return
    */
   @Override
   public String getDescription()
-- 
1.7.10.2