JAL-3200 fast lookup of DSSP3 notation

[jalview.git] / src / jalview / schemes / ResidueProperties.java

diff --git a/src/jalview/schemes/ResidueProperties.java b/src/jalview/schemes/ResidueProperties.java
index 55df1d1..18b11c1 100755 (executable)
--- a/src/jalview/schemes/ResidueProperties.java
+++ b/src/jalview/schemes/ResidueProperties.java
@@ -39,14 +39,14 @@ public class ResidueProperties
 
   public static final int[] purinepyrimidineIndex;
 
-  public static final Map<String, Integer> aa3Hash = new HashMap<String, Integer>();
+  public static final Map<String, Integer> aa3Hash = new HashMap<>();
 
-  public static final Map<String, String> aa2Triplet = new HashMap<String, String>();
+  public static final Map<String, String> aa2Triplet = new HashMap<>();
 
-  public static final Map<String, String> nucleotideName = new HashMap<String, String>();
+  public static final Map<String, String> nucleotideName = new HashMap<>();
 
   // lookup from modified amino acid (e.g. MSE) to canonical form (e.g. MET)
-  public static final Map<String, String> modifications = new HashMap<String, String>();
+  public static final Map<String, String> modifications = new HashMap<>();
 
   static
   {
@@ -496,25 +496,27 @@ public class ResidueProperties
    * Color.white, // R Color.white, // Y Color.white, // N Color.white, // Gap
    */
 
-  public static List<String> STOP = Arrays.asList("TGA", "TAA", "TAG");
+  public static String STOP = "STOP";
+
+  public static List<String> STOP_CODONS = Arrays.asList("TGA", "TAA", "TAG");
 
   public static String START = "ATG";
 
   /**
    * Nucleotide Ambiguity Codes
    */
-  public static final Map<String, String[]> ambiguityCodes = new Hashtable<String, String[]>();
+  public static final Map<String, String[]> ambiguityCodes = new Hashtable<>();
 
   /**
    * Codon triplets with additional symbols for unambiguous codons that include
    * ambiguity codes
    */
-  public static final Hashtable<String, String> codonHash2 = new Hashtable<String, String>();
+  public static final Hashtable<String, String> codonHash2 = new Hashtable<>();
 
   /**
    * all ambiguity codes for a given base
    */
-  public final static Hashtable<String, List<String>> _ambiguityCodes = new Hashtable<String, List<String>>();
+  public final static Hashtable<String, List<String>> _ambiguityCodes = new Hashtable<>();
 
   static
   {
@@ -638,7 +640,7 @@ public class ResidueProperties
         List<String> codesfor = _ambiguityCodes.get(r);
         if (codesfor == null)
         {
-          _ambiguityCodes.put(r, codesfor = new ArrayList<String>());
+          _ambiguityCodes.put(r, codesfor = new ArrayList<>());
         }
         if (!codesfor.contains(acode.getKey()))
         {
@@ -755,27 +757,27 @@ public class ResidueProperties
   }
 
   // Stores residue codes/names and colours and other things
-  public static Map<String, Map<String, Integer>> propHash = new Hashtable<String, Map<String, Integer>>();
+  public static Map<String, Map<String, Integer>> propHash = new Hashtable<>();
 
-  public static Map<String, Integer> hydrophobic = new Hashtable<String, Integer>();
+  public static Map<String, Integer> hydrophobic = new Hashtable<>();
 
-  public static Map<String, Integer> polar = new Hashtable<String, Integer>();
+  public static Map<String, Integer> polar = new Hashtable<>();
 
-  public static Map<String, Integer> small = new Hashtable<String, Integer>();
+  public static Map<String, Integer> small = new Hashtable<>();
 
-  public static Map<String, Integer> positive = new Hashtable<String, Integer>();
+  public static Map<String, Integer> positive = new Hashtable<>();
 
-  public static Map<String, Integer> negative = new Hashtable<String, Integer>();
+  public static Map<String, Integer> negative = new Hashtable<>();
 
-  public static Map<String, Integer> charged = new Hashtable<String, Integer>();
+  public static Map<String, Integer> charged = new Hashtable<>();
 
-  public static Map<String, Integer> aromatic = new Hashtable<String, Integer>();
+  public static Map<String, Integer> aromatic = new Hashtable<>();
 
-  public static Map<String, Integer> aliphatic = new Hashtable<String, Integer>();
+  public static Map<String, Integer> aliphatic = new Hashtable<>();
 
-  public static Map<String, Integer> tiny = new Hashtable<String, Integer>();
+  public static Map<String, Integer> tiny = new Hashtable<>();
 
-  public static Map<String, Integer> proline = new Hashtable<String, Integer>();
+  public static Map<String, Integer> proline = new Hashtable<>();
 
   static
   {
@@ -1149,52 +1151,56 @@ public class ResidueProperties
     String cdn = codonHash2.get(lccodon.toUpperCase());
     if ("*".equals(cdn))
     {
-      return "STOP";
+      return STOP;
     }
     return cdn;
   }
 
-  public static Hashtable<String, String> toDssp3State;
+  /*
+   * lookup of (A-Z) alternative secondary structure symbols'
+   * equivalents in DSSP3 notation
+   */
+  private static char[] toDssp3State;
   static
   {
-    toDssp3State = new Hashtable<String, String>();
-    toDssp3State.put("H", "H");
-    toDssp3State.put("E", "E");
-    toDssp3State.put("C", " ");
-    toDssp3State.put(" ", " ");
-    toDssp3State.put("T", " ");
-    toDssp3State.put("B", "E");
-    toDssp3State.put("G", "H");
-    toDssp3State.put("I", "H");
-    toDssp3State.put("X", " ");
+    toDssp3State = new char[9]; // for 'A'-'I'; extend if needed
+    Arrays.fill(toDssp3State, ' ');
+    toDssp3State['B' - 'A'] = 'E';
+    toDssp3State['E' - 'A'] = 'E';
+    toDssp3State['G' - 'A'] = 'H';
+    toDssp3State['H' - 'A'] = 'H';
+    toDssp3State['I' - 'A'] = 'H';
   }
 
   /**
    * translate from other dssp secondary structure alphabets to 3-state
    * 
-   * @param ssstring
-   * @return ssstring as a three-state secondary structure assignment.
+   * @param ssString
+   * @return ssstring
    */
-  public static String getDssp3state(String ssstring)
+  public static String getDssp3state(String ssString)
   {
-    if (ssstring == null)
+    if (ssString == null)
     {
       return null;
     }
-    StringBuffer ss = new StringBuffer();
-    for (int i = 0; i < ssstring.length(); i++)
+    int lookupSize = toDssp3State.length;
+    int len = ssString.length();
+    char[] trans = new char[len];
+    for (int i = 0; i < len; i++)
     {
-      String ssc = ssstring.substring(i, i + 1);
-      if (toDssp3State.containsKey(ssc))
+      char c = ssString.charAt(i);
+      int index = c - 'A';
+      if (index < 0 || index >= lookupSize)
       {
-        ss.append(toDssp3State.get(ssc));
+        trans[i] = ' ';
       }
       else
       {
-        ss.append(" ");
+        trans[i] = toDssp3State[index];
       }
     }
-    return ss.toString();
+    return new String(trans);
   }
 
   static
@@ -2525,7 +2531,7 @@ public class ResidueProperties
   // / cut here
   public static void main(String[] args)
   {
-    Hashtable<String, Vector<String>> aaProps = new Hashtable<String, Vector<String>>();
+    Hashtable<String, Vector<String>> aaProps = new Hashtable<>();
     System.out.println("my %aa = {");
     // invert property hashes
     for (String pname : propHash.keySet())
@@ -2536,7 +2542,7 @@ public class ResidueProperties
         Vector<String> aprops = aaProps.get(rname);
         if (aprops == null)
         {
-          aprops = new Vector<String>();
+          aprops = new Vector<>();
           aaProps.put(rname, aprops);
         }
         Integer hasprop = phash.get(rname);
@@ -2578,7 +2584,7 @@ public class ResidueProperties
   public static List<String> getResidues(boolean forNucleotide,
           boolean includeAmbiguous)
   {
-    List<String> result = new ArrayList<String>();
+    List<String> result = new ArrayList<>();
     if (forNucleotide)
     {
       for (String nuc : nucleotideName.keySet())