From: amwaterhouse <Andrew Waterhouse>
Date: Thu, 8 Dec 2005 09:18:08 +0000 (+0000)
Subject: After merge
X-Git-Tag: Release_2_07~39
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;h=58e82de6951f919828899de75b908cd1114d8511;p=jalview.git

After merge
---

diff --git a/src/jalview/analysis/SeqsetUtils.java b/src/jalview/analysis/SeqsetUtils.java
index 564e3f1..4432414 100755
--- a/src/jalview/analysis/SeqsetUtils.java
+++ b/src/jalview/analysis/SeqsetUtils.java
@@ -114,6 +114,14 @@ public class SeqsetUtils
     return new String("Sequence" + i);
   }
 
+  /**
+   * Generates a hash of SeqCharacterHash properties for each sequence
+   * in a sequence set, and optionally renames the sequences to an
+   * unambiguous 'safe' name.
+   * @param sequences SequenceI[]
+   * @param write_names boolean set this to rename each of the sequences to its unique_name(index) name
+   * @return Hashtable to be passed to @see deuniquify to recover original names (and properties) for renamed sequences
+   */
   public static Hashtable uniquify(SequenceI[] sequences, boolean write_names)
   {
     // Generate a safely named sequence set and a hash to recover the sequence names
@@ -134,25 +142,72 @@ public class SeqsetUtils
 
     return map;
   }
-
+  /**
+   * recover unsafe sequence names and original properties for a sequence
+   * set using a map generated by @see uniquify(sequences,true)
+   * @param map Hashtable
+   * @param sequences SequenceI[]
+   * @return boolean
+   */
   public static boolean deuniquify(Hashtable map, SequenceI[] sequences)
   {
-    // recover unsafe sequence names for a sequence set
-    boolean allfound = true;
-
-    for (int i = 0; i < sequences.length; i++)
-    {
-      if (map.containsKey(sequences[i].getName()))
-      {
-        Hashtable sqinfo = (Hashtable) map.get(sequences[i].getName());
-        SeqCharacterUnhash(sequences[i], sqinfo);
-      }
-      else
-      {
-        allfound = false;
+    jalview.analysis.SequenceIdMatcher matcher = new SequenceIdMatcher(sequences);
+    SequenceI msq = null;
+    Enumeration keys = map.keys();
+    Vector unmatched = new Vector();
+    for (int i=0, j=sequences.length; i<j; i++)
+      unmatched.add(sequences[i]);
+    while (keys.hasMoreElements()) {
+      Object key = keys.nextElement();
+      if (key instanceof String) {
+        if ((msq = matcher.findIdMatch((String) key))!=null) {
+          Hashtable sqinfo = (Hashtable) map.get(key);
+          unmatched.remove(msq);
+          SeqCharacterUnhash(msq, sqinfo);
+        }
+        else
+        {
+          System.err.println("Can't find '"+((String) key)+"' in uniquified alignment");
+        }
       }
     }
+    if (unmatched.size()>0) {
+      System.err.println("Did not find matches for :");
+      for (Enumeration i = unmatched.elements(); i.hasMoreElements(); System.out.println(((SequenceI) i.nextElement()).getName()))
+           ;
+      return false;
+    }
 
-    return allfound;
+    return true;
+  }
+  /**
+   * returns a subset of the sequenceI seuqences,
+   * including only those that contain at least one residue.
+   * @param sequences SequenceI[]
+   * @return SequenceI[]
+   */
+  public static SequenceI[] getNonEmptySequenceSet(SequenceI[] sequences) {
+      // Identify first row of alignment with residues for prediction
+      boolean ungapped[] = new boolean[sequences.length];
+      int msflen=0;
+      for (int i=0,j=sequences.length; i<j;i++) {
+        String tempseq = jalview.analysis.AlignSeq.extractGaps(jalview.util.Comparison.GapChars, sequences[i].getSequence());
+        if (tempseq.length()==0)
+          ungapped[i]=false;
+        else {
+          ungapped[i]=true;
+          msflen++;
+        }
+      }
+      if (msflen==0)
+        return null; // no minimal set
+      // compose minimal set
+      SequenceI[] mset = new SequenceI[msflen];
+      for (int i=0,j=sequences.length,k=0; i<j;i++) {
+        if (ungapped[i])
+          mset[k++] = sequences[i];
+      }
+      ungapped = null;
+      return mset;
   }
 }
diff --git a/src/jalview/ws/JPredClient.java b/src/jalview/ws/JPredClient.java
index 4cb8aee..f503f91 100755
--- a/src/jalview/ws/JPredClient.java
+++ b/src/jalview/ws/JPredClient.java
@@ -57,6 +57,7 @@ public class JPredClient
       wsInfo = setWebService();
 
     SequenceI seq = msf[0];
+
     altitle = "JNet prediction on " + seq.getName() +
         " using alignment from " + title;
 
@@ -65,8 +66,10 @@ public class JPredClient
                            AlignSeq.extractGaps("-. ", seq.getSequence()) +
                            "\n");
     SequenceI aln[] = new SequenceI[msf.length];
-    for (int i=0,j=msf.length; i<j;i++)
-      aln[i] = new jalview.datamodel.Sequence(msf[i]);
+    for (int i=0,j=msf.length; i<j;i++) {
+        aln[i] = new jalview.datamodel.Sequence(msf[i]);
+    }
+
     SequenceInfo = jalview.analysis.SeqsetUtils.uniquify(aln, true);
 
     if (!locateWebService())
@@ -403,16 +406,17 @@ public class JPredClient
           {
             al = new Alignment(new FormatAdapter().readFile(
                 result.getAligfile(), "Paste", format));
+            SequenceI sqs[] = new SequenceI[al.getHeight()];
             for (int i=0, j=al.getHeight(); i<j; i++) {
-              SequenceI sq = al.getSequenceAt(i);
-              if (!jalview.analysis.SeqsetUtils.SeqCharacterUnhash(
-                  sq, (Hashtable) SequenceInfo.get(sq.getName())))
+              sqs[i] = al.getSequenceAt(i);
+            }
+            if (!jalview.analysis.SeqsetUtils.deuniquify(
+                  (Hashtable) SequenceInfo,sqs))
               {
                 throw (new Exception(
-                    "Couldn't recover sequence properties for JNet "
-                    +((i==0) ? "Query sequence" : "alignment sequence ("+i+")")));
+                    "Couldn't recover sequence properties for alignment."));
               }
-            }
+
             noMsa = false;
             FirstSeq = 0;
           }