Merge branch 'releases/Release_2_11_3_Branch'

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

diff --git a/src/jalview/util/Comparison.java b/src/jalview/util/Comparison.java
index af6052c..e9ff931 100644 (file)
--- a/src/jalview/util/Comparison.java
+++ b/src/jalview/util/Comparison.java
@@ -42,6 +42,8 @@ public class Comparison
 
   private static final int NUCLEOTIDE_COUNT_VERY_SHORT_SEQUENCE;
 
+  private static final boolean NUCLEOTIDE_AMBIGUITY_DETECTION;
+
   public static final char GAP_SPACE = ' ';
 
   public static final char GAP_DOT = '.';
@@ -54,6 +56,7 @@ public class Comparison
 
   static
   {
+    // these options read only at start of session
     NUCLEOTIDE_COUNT_PERCENT = Cache.getDefault("NUCLEOTIDE_COUNT_PERCENT",
             55);
     NUCLEOTIDE_COUNT_LONG_SEQUENCE_AMBIGUITY_PERCENT = Cache.getDefault(
@@ -62,6 +65,8 @@ public class Comparison
             .getDefault("NUCLEOTIDE_COUNT_SHORT", 100);
     NUCLEOTIDE_COUNT_VERY_SHORT_SEQUENCE = Cache
             .getDefault("NUCLEOTIDE_COUNT_VERY_SHORT", 4);
+    NUCLEOTIDE_AMBIGUITY_DETECTION = Cache
+            .getDefault("NUCLEOTIDE_AMBIGUITY_DETECTION", true);
   }
 
   /**
@@ -324,17 +329,17 @@ public class Comparison
     }
     long allCount = ntCount + aaCount;
 
-    if (Cache.getDefault("NUCLEOTIDE_AMBIGUITY_DETECTION", true))
+    if (NUCLEOTIDE_AMBIGUITY_DETECTION)
     {
-      Console.info("Performing new nucleotide detection routine");
+      Console.debug("Performing new nucleotide detection routine");
       if (allCount > NUCLEOTIDE_COUNT_SHORT_SEQUENCE)
       {
         // a long sequence.
         // check for at least 55% nucleotide, and nucleotide and ambiguity codes
         // (including N) must make up 95%
-        return ntCount * 100 > NUCLEOTIDE_COUNT_PERCENT * allCount
+        return ntCount * 100 >= NUCLEOTIDE_COUNT_PERCENT * allCount
                 && 100 * (ntCount + nCount
-                        + ntaCount) > NUCLEOTIDE_COUNT_LONG_SEQUENCE_AMBIGUITY_PERCENT
+                        + ntaCount) >= NUCLEOTIDE_COUNT_LONG_SEQUENCE_AMBIGUITY_PERCENT
                                 * allCount;
       }
       else if (allCount > NUCLEOTIDE_COUNT_VERY_SHORT_SEQUENCE)
@@ -342,7 +347,7 @@ public class Comparison
         // a short sequence.
         // check if a short sequence is at least 55% nucleotide and the rest of
         // the symbols are all X or all N
-        if (ntCount * 100 > NUCLEOTIDE_COUNT_PERCENT * allCount
+        if (ntCount * 100 >= NUCLEOTIDE_COUNT_PERCENT * allCount
                 && (nCount == aaCount || xCount == aaCount))
         {
           return true;
@@ -365,7 +370,7 @@ public class Comparison
     }
     else
     {
-      Console.info("Performing old nucleotide detection routine");
+      Console.debug("Performing old nucleotide detection routine");
       /*
        * Check for nucleotide count > 85% of total count (in a form that evades
        * int / float conversion or divide by zero).
@@ -393,18 +398,19 @@ public class Comparison
     // vs := NUCLEOTIDE_COUNT_VERY_SHORT_SEQUENCE
     // s := NUCLEOTIDE_COUNT_SHORT_SEQUENCE
     // lp := NUCLEOTIDE_COUNT_LOWER_PERCENT
-    // vsp := 1 - (1/a) # this is the proportion of required nucleotides in
-    // # a VERY_SHORT Sequence (4 bases).
-    // # should be all but one base is nucleotide.
+    // vsp := 1 - (1/a) # this is the proportion of required definite
+    // nucleotides
+    // # in a VERY_SHORT Sequence (4 bases).
+    // # This should be equivalent to all but one base in the sequence.
     // p := (a - vs)/(s - vs) # proportion of the way between
     // # VERY_SHORT and SHORT thresholds.
-    // tp := vsp + p * (lp/100 - vsp) # the proportion of nucleotides
+    // tp := vsp + p * (lp/100 - vsp) # the proportion of definite nucleotides
     // # required for this length of sequence.
     // minNt := tp * a # the minimum number of definite nucleotide bases
-    // # required for this length of sequences.
+    // # required for this length of sequence.
     //
     // We are then essentially returning:
-    // # ntCount >= 55% of allCount and the rest are all nucleotide ambiguity
+    // # ntCount >= 55% of allCount and the rest are all nucleotide ambiguity:
     // ntCount >= tp * allCount && nCount + ntaCount == aaCount
     // but without going into float/double land
     long LHS = 100 * allCount