JAL-2214 JAL-1841 compute canonical, canonicalAndWobble visually distinguish dominant...

[jalview.git] / src / jalview / analysis / StructureFrequency.java

diff --git a/src/jalview/analysis/StructureFrequency.java b/src/jalview/analysis/StructureFrequency.java
index 29d02fe..ad3f8d9 100644 (file)
--- a/src/jalview/analysis/StructureFrequency.java
+++ b/src/jalview/analysis/StructureFrequency.java
@@ -24,6 +24,7 @@ import jalview.datamodel.AlignmentAnnotation;
 import jalview.datamodel.Annotation;
 import jalview.datamodel.SequenceFeature;
 import jalview.datamodel.SequenceI;
+import jalview.util.Comparison;
 import jalview.util.Format;
 
 import java.util.ArrayList;
@@ -111,8 +112,9 @@ public class StructureFrequency
 
     for (int i = start; i < end; i++) // foreach column
     {
-      int canonicalOrWobblePairCount = 0;
+      int canonicalOrWobblePairCount = 0, canonical = 0;
       int otherPairCount = 0;
+      int nongap = 0;
       maxResidue = "-";
       values = new int[255];
       pairs = new int[255][255];
@@ -151,21 +153,16 @@ public class StructureFrequency
                       .println("WARNING: Consensus skipping null sequence - possible race condition.");
               continue;
             }
-            c = sequences[j].getCharAt(i);
 
-            // standard representation for gaps in sequence and structure
-            if (c == '.' || c == ' ')
-            {
-              c = '-';
-            }
+            c = sequences[j].getCharAt(i);
+            cEnd = sequences[j].getCharAt(bpEnd);
 
-            if (c == '-')
+            if (Comparison.isGap(c) || Comparison.isGap(cEnd))
             {
               values['-']++;
               continue;
             }
-            cEnd = sequences[j].getCharAt(bpEnd);
-
+            nongap++;
             /*
              * ensure upper-case for counting purposes
              */
@@ -179,20 +176,19 @@ public class StructureFrequency
             }
             if (Rna.isCanonicalOrWobblePair(c, cEnd))
             {
-              values['(']++;
-              maxResidue = "(";
               canonicalOrWobblePairCount++;
+              if (Rna.isCanonicalPair(c, cEnd))
+              {
+                canonical++;
+              }
             }
             else
             {
-              values['[']++;
-              maxResidue = "[";
               otherPairCount++;
             }
             pairs[c][cEnd]++;
           }
         }
-        // nonGap++;
       }
 
       residueHash = new Hashtable();
@@ -204,11 +200,30 @@ public class StructureFrequency
 
         residueHash.put(PAIRPROFILE, pairs);
       }
-      int count = Math.max(canonicalOrWobblePairCount, otherPairCount);
-      if (!maxResidue.equals("-"))
+      values['('] = canonicalOrWobblePairCount;
+      values['['] = canonical;
+      values['{'] = otherPairCount;
+      /*
+       * the count is the number of valid pairs (as a percentage, determines
+       * the relative size of the profile logo)
+       */
+      int count = canonicalOrWobblePairCount;
+
+      /*
+       * display '(' if most pairs are canonical, or as
+       * '[' if there are more wobble pairs. 
+       */
+      if (canonicalOrWobblePairCount > 0 || otherPairCount > 0)
       {
-        maxResidue = canonicalOrWobblePairCount >= otherPairCount ? "("
+        if (canonicalOrWobblePairCount >= otherPairCount)
+        {
+          maxResidue = (canonicalOrWobblePairCount - canonical) < canonical ? "("
                 : "[";
+        }
+        else
+        {
+          maxResidue = "{";
+        }
       }
       residueHash.put(MAXCOUNT, new Integer(count));
       residueHash.put(MAXRESIDUE, maxResidue);
@@ -216,8 +231,9 @@ public class StructureFrequency
       percentage = ((float) count * 100) / jSize;
       residueHash.put(PID_GAPS, new Float(percentage));
 
-      // percentage = ((float) count * 100) / (float) nongap;
-      // residueHash.put(PID_NOGAPS, new Float(percentage));
+      percentage = ((float) count * 100) / nongap;
+      residueHash.put(PID_NOGAPS, new Float(percentage));
+
       if (result[i] == null)
       {
         result[i] = residueHash;
@@ -226,9 +242,12 @@ public class StructureFrequency
       {
         values[')'] = values['('];
         values[']'] = values['['];
+        values['}'] = values['{'];
         values['('] = 0;
         values['['] = 0;
-        maxResidue = maxResidue.equals("(") ? ")" : "]";
+        values['{'] = 0;
+        maxResidue = maxResidue.equals("(") ? ")"
+                : maxResidue.equals("[") ? "]" : "}";
 
         residueHash = new Hashtable();
         if (profile)
@@ -245,6 +264,9 @@ public class StructureFrequency
         percentage = ((float) count * 100) / jSize;
         residueHash.put(PID_GAPS, new Float(percentage));
 
+        percentage = ((float) count * 100) / nongap;
+        residueHash.put(PID_NOGAPS, new Float(percentage));
+
         result[bpEnd] = residueHash;
       }
     }