Merge branch 'alpha/JAL-3362_Jalview_212_alpha' into alpha/merge_212_JalviewJS_2112

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

diff --git a/src/jalview/schemes/ResidueProperties.java b/src/jalview/schemes/ResidueProperties.java
index bf30ed6..0b62bb6 100755 (executable)
--- a/src/jalview/schemes/ResidueProperties.java
+++ b/src/jalview/schemes/ResidueProperties.java
@@ -34,6 +34,13 @@ import java.util.Vector;
 
 public class ResidueProperties
 {
+  // alphabet names used in Hidden Markov Model files
+  public static final String ALPHABET_RNA = "RNA";
+
+  public static final String ALPHABET_DNA = "DNA";
+
+  public static final String ALPHABET_AMINO = "amino";
+
   // Stores residue codes/names and colours and other things
   public static final int[] aaIndex; // aaHash version 2.1.1 and below
 
@@ -50,6 +57,9 @@ public class ResidueProperties
   // lookup from modified amino acid (e.g. MSE) to canonical form (e.g. MET)
   public static final Map<String, String> modifications = new HashMap<>();
 
+  // residue background frequencies across different alphabets
+  public static final Map<String, Map<Character, Float>> backgroundFrequencies = new HashMap<>();
+
   static
   {
     aaIndex = new int[255];
@@ -905,47 +915,51 @@ public class ResidueProperties
     return peptide;
   }
 
-  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<>();
-    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
@@ -2266,6 +2280,58 @@ public class ResidueProperties
 
   }
 
+  static
+  {
+    Map<Character, Float> amino = new HashMap<>();
+    amino.put('A', 0.0826f);
+    amino.put('Q', 0.0393f);
+    amino.put('L', 0.0965f);
+    amino.put('S', 0.0661f);
+    amino.put('R', 0.0553f);
+    amino.put('E', 0.0674f);
+    amino.put('K', 0.0582f);
+    amino.put('T', 0.0535f);
+    amino.put('N', 0.0406f);
+    amino.put('G', 0.0708f);
+    amino.put('M', 0.0241f);
+    amino.put('W', 0.0109f);
+    amino.put('D', 0.0546f);
+    amino.put('H', 0.0227f);
+    amino.put('F', 0.0386f);
+    amino.put('Y', 0.0292f);
+    amino.put('C', 0.0137f);
+    amino.put('I', 0.0593f);
+    amino.put('P', 0.0472f);
+    amino.put('V', 0.0686f);
+    backgroundFrequencies.put(ALPHABET_AMINO, amino);
+    // todo: these don't match https://www.ebi.ac.uk/uniprot/TrEMBLstats - what
+    // are they?
+  }
+
+  // TODO get correct frequencies
+
+  static
+  {
+    Map<Character, Float> dna = new HashMap<>();
+    dna.put('A', 0.25f);
+    dna.put('C', 0.25f);
+    dna.put('T', 0.25f);
+    dna.put('G', 0.25f);
+    backgroundFrequencies.put(ALPHABET_DNA, dna);
+
+  }
+
+  static
+  {
+    Map<Character, Float> rna = new HashMap<>();
+    rna.put('A', 0.25f);
+    rna.put('C', 0.25f);
+    rna.put('T', 0.25f);
+    rna.put('G', 0.25f);
+    backgroundFrequencies.put(ALPHABET_RNA, rna);
+
+  }
+
   public static String getCanonicalAminoAcid(String aA)
   {
     String canonical = modifications.get(aA);
@@ -2274,6 +2340,10 @@ public class ResidueProperties
 
   // main method generates perl representation of residue property hash
   // / cut here
+  /**
+   * @j2sIgnore
+   * @param args
+   */
   public static void main(String[] args)
   {
     Hashtable<String, Vector<String>> aaProps = new Hashtable<>();