From: Jim Procter <jprocter@issues.jalview.org>
Date: Tue, 29 Sep 2020 14:26:34 +0000 (+0100)
Subject: JAL-3760 refactor nucleotide proportion test to per-sequence and return false if... 
X-Git-Tag: Release_2_11_1_3~7^2~1
X-Git-Url: http://source.jalview.org/gitweb/?p=jalview.git;a=commitdiff_plain;h=75ea8cab6b3f375b65a6146c7de37f4107294cc5

JAL-3760 refactor nucleotide proportion test to per-sequence and return false if any sequence appears protein-like
---

diff --git a/src/jalview/util/Comparison.java b/src/jalview/util/Comparison.java
index d4fc233..286bfb2 100644
--- a/src/jalview/util/Comparison.java
+++ b/src/jalview/util/Comparison.java
@@ -268,7 +268,38 @@ public class Comparison
    */
   public static final boolean isNucleotide(SequenceI seq)
   {
-    return isNucleotide(new SequenceI[] { seq });
+    if (seq==null)
+    {
+      return false;
+    }
+    long ntCount = 0;
+    long aaCount = 0;
+
+    int len = seq.getLength();
+    for (int i = 0; i < len; i++)
+    {
+      char c = seq.getCharAt(i);
+      if (isNucleotide(c))
+      {
+        ntCount++;
+      }
+      else if (!isGap(c))
+      {
+        aaCount++;
+      }
+    }
+    /*
+     * Check for nucleotide count > 85% of total count (in a form that evades
+     * int / float conversion or divide by zero).
+     */
+    if (ntCount * 100 > EIGHTY_FIVE * (ntCount + aaCount))
+    {
+      return true;
+    }
+    else
+    {
+      return false;
+    }
   }
 
   /**
@@ -285,45 +316,23 @@ public class Comparison
     {
       return false;
     }
-
-    int ntCount = 0;
-    int aaCount = 0;
+    // true if we have seen a nucleotide sequence
+    boolean na=false;
     for (SequenceI seq : seqs)
     {
       if (seq == null)
       {
         continue;
       }
+      na=true;
       // TODO could possibly make an informed guess just from the first sequence
       // to save a lengthy calculation
-      int len = seq.getLength();
-      for (int i = 0; i < len; i++)
-      {
-        char c = seq.getCharAt(i);
-        if (isNucleotide(c))
-        {
-          ntCount++;
-        }
-        else if (!isGap(c))
-        {
-          aaCount++;
-        }
+      if (seq.isProtein()) {
+        // if even one looks like protein, the alignment is protein
+        return false;
       }
     }
-
-    /*
-     * Check for nucleotide count > 85% of total count (in a form that evades
-     * int / float conversion or divide by zero).
-     */
-    if (ntCount * 100 > EIGHTY_FIVE * (ntCount + aaCount))
-    {
-      return true;
-    }
-    else
-    {
-      return false;
-    }
-
+    return na;
   }
 
   /**