JAL-1705 refactored/utility methods to detect e.g. 'PhenCode_variation'

[jalview.git] / src / jalview / util / Comparison.java

diff --git a/src/jalview/util/Comparison.java b/src/jalview/util/Comparison.java
index dfc243d..5605a53 100644 (file)
--- a/src/jalview/util/Comparison.java
+++ b/src/jalview/util/Comparison.java
@@ -265,10 +265,13 @@ public class Comparison
     char[][] letters = new char[seqs.length][];
     for (int i = 0; i < seqs.length; i++)
     {
-      char[] sequence = seqs[i].getSequence();
-      if (sequence != null)
+      if (seqs[i] != null)
       {
-        letters[i] = sequence;
+        char[] sequence = seqs[i].getSequence();
+        if (sequence != null)
+        {
+          letters[i] = sequence;
+        }
       }
     }
 
@@ -283,7 +286,7 @@ public class Comparison
    * @param letters
    * @return
    */
-  public static final boolean areNucleotide(char[][] letters)
+  static final boolean areNucleotide(char[][] letters)
   {
     int ntCount = 0;
     int aaCount = 0;
@@ -297,16 +300,11 @@ public class Comparison
       // to save a lengthy calculation
       for (char c : seq)
       {
-        if ('a' <= c && c <= 'z')
-        {
-          c -= TO_UPPER_CASE;
-        }
-
-        if (c == 'A' || c == 'G' || c == 'C' || c == 'T' || c == 'U')
+        if (isNucleotide(c))
         {
           ntCount++;
         }
-        else if (!Comparison.isGap(c))
+        else if (!isGap(c))
         {
           aaCount++;
         }
@@ -329,6 +327,59 @@ public class Comparison
   }
 
   /**
+   * Answers true if the character is one of aAcCgGtTuU
+   * 
+   * @param c
+   * @return
+   */
+  public static boolean isNucleotide(char c)
+  {
+    if ('a' <= c && c <= 'z')
+    {
+      c -= TO_UPPER_CASE;
+    }
+
+    switch (c)
+    {
+    case 'A':
+    case 'C':
+    case 'G':
+    case 'T':
+    case 'U':
+      return true;
+    }
+    return false;
+  }
+
+  /**
+   * Answers true if every character in the string is one of aAcCgGtTuU, or
+   * (optionally) a gap character (dot, dash, space), else false
+   * 
+   * @param s
+   * @param allowGaps
+   * @return
+   */
+  public static boolean isNucleotideSequence(String s, boolean allowGaps)
+  {
+    if (s == null)
+    {
+      return false;
+    }
+    for (int i = 0; i < s.length(); i++)
+    {
+      char c = s.charAt(i);
+      if (!isNucleotide(c))
+      {
+        if (!allowGaps || !isGap(c))
+        {
+          return false;
+        }
+      }
+    }
+    return true;
+  }
+
+  /**
    * Convenience overload of isNucleotide
    * 
    * @param seqs