From cb37b13542829ad0406c5f0ff8349334f83c31ae Mon Sep 17 00:00:00 2001
From: jprocter <Jim Procter>
Date: Fri, 19 Jun 2009 10:01:09 +0000
Subject: [PATCH] bugfix mapped reference transfer and made picr ID retrieval
 optional

---
 src/jalview/ws/DBRefFetcher.java |  172 +++++++++++++++++++++++++-------------
 1 file changed, 112 insertions(+), 60 deletions(-)

diff --git a/src/jalview/ws/DBRefFetcher.java b/src/jalview/ws/DBRefFetcher.java
index 4b7c07e..1fd9f1c 100644
--- a/src/jalview/ws/DBRefFetcher.java
+++ b/src/jalview/ws/DBRefFetcher.java
@@ -19,6 +19,7 @@
 package jalview.ws;
 
 import jalview.analysis.AlignSeq;
+import jalview.bin.Cache;
 import jalview.datamodel.AlignmentI;
 import jalview.datamodel.DBRefEntry;
 import jalview.datamodel.DBRefSource;
@@ -59,6 +60,7 @@ public class DBRefFetcher implements Runnable
   StringBuffer sbuffer = new StringBuffer();
 
   boolean running = false;
+
   /**
    * picr client instance
    */
@@ -77,27 +79,29 @@ public class DBRefFetcher implements Runnable
   }
 
   /**
-   * Creates a new SequenceFeatureFetcher object and fetches from the
-   * currently selected set of databases.
+   * Creates a new SequenceFeatureFetcher object and fetches from the currently
+   * selected set of databases.
    * 
    * @param seqs
-   *                fetch references for these sequences
+   *          fetch references for these sequences
    * @param af
-   *                the parent alignframe for progress bar monitoring.
+   *          the parent alignframe for progress bar monitoring.
    */
   public DBRefFetcher(SequenceI[] seqs, AlignFrame af)
   {
     this(seqs, af, null);
   }
+
   /**
-   * Creates a new SequenceFeatureFetcher object and fetches from the
-   * currently selected set of databases.
+   * Creates a new SequenceFeatureFetcher object and fetches from the currently
+   * selected set of databases.
    * 
    * @param seqs
-   *                fetch references for these sequences
+   *          fetch references for these sequences
    * @param af
-   *                the parent alignframe for progress bar monitoring.
-   * @param sources array of database source strings to query references from
+   *          the parent alignframe for progress bar monitoring.
+   * @param sources
+   *          array of database source strings to query references from
    */
   public DBRefFetcher(SequenceI[] seqs, AlignFrame af, String[] sources)
   {
@@ -113,11 +117,13 @@ public class DBRefFetcher implements Runnable
     this.dataset = ds;
     // TODO Jalview 2.5 lots of this code should be in the gui package!
     sfetcher = jalview.gui.SequenceFetcher.getSequenceFetcherSingleton(af);
-    if (sources==null)
+    if (sources == null)
     {
       // af.featureSettings_actionPerformed(null);
-      String[] defdb=null,otherdb = sfetcher.getDbInstances(jalview.ws.dbsources.DasSequenceSource.class);
-      Vector selsources = new Vector(), dasselsrc= (af.featureSettings!=null) ? af.featureSettings.getSelectedSources()
+      String[] defdb = null, otherdb = sfetcher
+              .getDbInstances(jalview.ws.dbsources.DasSequenceSource.class);
+      Vector selsources = new Vector(), dasselsrc = (af.featureSettings != null) ? af.featureSettings
+              .getSelectedSources()
               : new jalview.gui.DasSourceBrowser().getSelectedSources();
       Enumeration en = dasselsrc.elements();
       while (en.hasMoreElements())
@@ -126,12 +132,14 @@ public class DBRefFetcher implements Runnable
         selsources.addElement(src.getNickname());
       }
       int osel = 0;
-      for (int o=0;otherdb!=null && o<otherdb.length;o++)
+      for (int o = 0; otherdb != null && o < otherdb.length; o++)
       {
         if (!selsources.contains(otherdb[o]))
         {
           otherdb[o] = null;
-        } else {
+        }
+        else
+        {
           osel++;
         }
       }
@@ -144,45 +152,56 @@ public class DBRefFetcher implements Runnable
       {
         defdb = DBRefSource.PROTEINDBS;
       }
-      // append the selected sequence sources to the default dbs 
-      dbSources = new String[defdb.length+osel];
+      // append the selected sequence sources to the default dbs
+      dbSources = new String[defdb.length + osel];
       System.arraycopy(defdb, 0, dbSources, 0, defdb.length);
-      for (int o=0,op=defdb.length; otherdb!=null && o<otherdb.length; o++)
+      for (int o = 0, op = defdb.length; otherdb != null
+              && o < otherdb.length; o++)
       {
-        if (otherdb[o]!=null)
+        if (otherdb[o] != null)
         {
           dbSources[op++] = otherdb[o];
         }
       }
-    } else {
-      // we assume the caller knows what they're doing and ensured that all the db source names are valid
+    }
+    else
+    {
+      // we assume the caller knows what they're doing and ensured that all the
+      // db source names are valid
       dbSources = sources;
     }
   }
+
   /**
-   * retrieve all the das sequence sources and add them to the list of db sources to retrieve from 
+   * retrieve all the das sequence sources and add them to the list of db
+   * sources to retrieve from
    */
   public void appendAllDasSources()
   {
     if (dbSources == null)
     {
-      dbSources = new String[] {};
+      dbSources = new String[]
+      {};
     }
     // append additional sources
-    String[] otherdb = sfetcher.getDbInstances(jalview.ws.dbsources.DasSequenceSource.class);
-    if (otherdb!=null && otherdb.length>0)
+    String[] otherdb = sfetcher
+            .getDbInstances(jalview.ws.dbsources.DasSequenceSource.class);
+    if (otherdb != null && otherdb.length > 0)
     {
-      String[] newsrc = new String[dbSources.length+otherdb.length];
-      System.arraycopy(dbSources, 0, newsrc,0,dbSources.length);
-      System.arraycopy(otherdb, 0, newsrc,dbSources.length, otherdb.length);
+      String[] newsrc = new String[dbSources.length + otherdb.length];
+      System.arraycopy(dbSources, 0, newsrc, 0, dbSources.length);
+      System
+              .arraycopy(otherdb, 0, newsrc, dbSources.length,
+                      otherdb.length);
       dbSources = newsrc;
     }
   }
+
   /**
    * start the fetcher thread
    * 
    * @param waitTillFinished
-   *                true to block until the fetcher has finished
+   *          true to block until the fetcher has finished
    */
   public void fetchDBRefs(boolean waitTillFinished)
   {
@@ -209,9 +228,9 @@ public class DBRefFetcher implements Runnable
    * could be either seq name or dbref id
    * 
    * @param seq
-   *                SequenceI
+   *          SequenceI
    * @param key
-   *                String
+   *          String
    */
   void addSeqId(SequenceI seq, String key)
   {
@@ -254,8 +273,13 @@ public class DBRefFetcher implements Runnable
     running = true;
     long startTime = System.currentTimeMillis();
     af.setProgressBar("Fetching db refs", startTime);
-    try {
-      picrClient = new uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperServiceLocator().getAccessionMapperPort(); 
+    try
+    {
+      if (Cache.getDefault("DBREFFETCH_USEPICR", false))
+      {
+        picrClient = new uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperServiceLocator()
+                .getAccessionMapperPort();
+      }
     } catch (Exception e)
     {
       System.err.println("Couldn't locate PICR service instance.\n");
@@ -304,10 +328,10 @@ public class DBRefFetcher implements Runnable
         {
           // Still queries to make for current seqIndex
           StringBuffer queryString = new StringBuffer("");
-          int numq=0,nqSize = (maxqlen > queries.size()) ? queries.size()
-                  : maxqlen;
-          
-          while (queries.size()>0 && numq < nqSize)
+          int numq = 0, nqSize = (maxqlen > queries.size()) ? queries
+                  .size() : maxqlen;
+
+          while (queries.size() > 0 && numq < nqSize)
           {
             String query = (String) queries.elementAt(0);
             if (dbsource.isValidReference(query))
@@ -326,7 +350,9 @@ public class DBRefFetcher implements Runnable
           {
             if (jalview.bin.Cache.log.isDebugEnabled())
             {
-              jalview.bin.Cache.log.debug("Querying "+dbsource.getDbName()+" with : '"+queryString.toString()+"'");
+              jalview.bin.Cache.log.debug("Querying "
+                      + dbsource.getDbName() + " with : '"
+                      + queryString.toString() + "'");
             }
             retrieved = dbsource.getSequenceRecords(queryString.toString());
           } catch (Exception ex)
@@ -353,7 +379,7 @@ public class DBRefFetcher implements Runnable
                     { dbSources[db] }); // jalview.datamodel.DBRefSource.UNIPROT
             // });
             // check for existing dbrefs to use
-            if (uprefs != null && uprefs.length>0)
+            if (uprefs != null && uprefs.length > 0)
             {
               for (int j = 0; j < uprefs.length; j++)
               {
@@ -372,25 +398,39 @@ public class DBRefFetcher implements Runnable
               {
                 String token = st.nextToken();
                 UPEntry[] presp = null;
-                try {
-                  presp=picrClient.getUPIForAccession(token, null, picrClient.getMappedDatabaseNames(), null, true);
-                } catch (Exception e) {
-                  System.err.println("Exception with Picr for '"+token+"'\n");
-                  e.printStackTrace();
+                if (picrClient != null)
+                {
+                  // resolve the string against PICR to recover valid IDs
+                  try
+                  {
+                    presp = picrClient
+                            .getUPIForAccession(token, null, picrClient
+                                    .getMappedDatabaseNames(), null, true);
+                  } catch (Exception e)
+                  {
+                    System.err.println("Exception with Picr for '" + token
+                            + "'\n");
+                    e.printStackTrace();
+                  }
                 }
-                if (presp!=null && presp.length>0)
+                if (presp != null && presp.length > 0)
                 {
-                  for (int id=0;id<presp.length; id++)
+                  for (int id = 0; id < presp.length; id++)
                   {
-                    // construct sequences from response if sequences are present, and do a transferReferences
+                    // construct sequences from response if sequences are
+                    // present, and do a transferReferences
                     // otherwise transfer non sequence x-references directly.
                   }
-                  System.out.println("Validated ID against PICR... (for what its worth):"+token);
+                  System.out
+                          .println("Validated ID against PICR... (for what its worth):"
+                                  + token);
                   addSeqId(sequence, token);
                   queries.addElement(token.toUpperCase());
-                } else {
-                  //if ()
-                  //System.out.println("Not querying source with token="+token+"\n");
+                }
+                else
+                {
+                  // if ()
+                  // System.out.println("Not querying source with token="+token+"\n");
                   addSeqId(sequence, token);
                   queries.addElement(token.toUpperCase());
                 }
@@ -431,13 +471,15 @@ public class DBRefFetcher implements Runnable
           AlignmentI retrievedAl) // File
   // file)
   {
-
     if (retrievedAl == null || retrievedAl.getHeight() == 0)
     {
       return;
     }
-    SequenceI[] retrieved = recoverDbSequences(retrievedAl.getSequencesArray());
+    SequenceI[] retrieved = recoverDbSequences(retrievedAl
+            .getSequencesArray());
     SequenceI sequence = null;
+    boolean transferred = false;
+    StringBuffer messages = new StringBuffer();
 
     // Vector entries = new Uniprot().getUniprotEntries(file);
 
@@ -530,11 +572,11 @@ public class DBRefFetcher implements Runnable
           absStart = nonGapped.indexOf(entrySeq);
           if (absStart == -1)
           { // verification failed.
-            sbuffer.append(sequence.getName()
+            messages.append(sequence.getName()
                     + " SEQUENCE NOT %100 MATCH \n");
             continue;
           }
-
+          transferred = true;
           sbuffer.append(sequence.getName() + " HAS " + absStart
                   + " PREFIXED RESIDUES COMPARED TO " + dbSource + "\n");
           //
@@ -554,6 +596,7 @@ public class DBRefFetcher implements Runnable
         }
         else
         {
+          transferred = true;
           // update start and end of local sequence to place it in entry's
           // reference frame.
           // apply identity map map from whole of local sequence to matching
@@ -599,35 +642,44 @@ public class DBRefFetcher implements Runnable
         // and remove it from the rest
         // TODO: decide if we should remove annotated sequence from set
         sdataset.remove(sequence);
-        // TODO: should we make a note of sequences that have received new DB ids, so we can query all enabled DAS servers for them ?
+        // TODO: should we make a note of sequences that have received new DB
+        // ids, so we can query all enabled DAS servers for them ?
       }
     }
+    if (!transferred)
+    {
+      // report the ID/sequence mismatches
+      sbuffer.append(messages);
+    }
   }
 
   /**
    * loop thru and collect additional sequences in Map.
+   * 
    * @param sequencesArray
    * @return
    */
   private SequenceI[] recoverDbSequences(SequenceI[] sequencesArray)
   {
     Vector nseq = new Vector();
-    for (int i=0;sequencesArray!=null && i<sequencesArray.length;i++)
+    for (int i = 0; sequencesArray != null && i < sequencesArray.length; i++)
     {
       nseq.addElement(sequencesArray[i]);
       DBRefEntry dbr[] = sequencesArray[i].getDBRef();
       jalview.datamodel.Mapping map = null;
-      for (int r=0;(dbr!=null) && r<dbr.length; r++) {
-        if ((map=dbr[r].getMap())!=null)
+      for (int r = 0; (dbr != null) && r < dbr.length; r++)
+      {
+        if ((map = dbr[r].getMap()) != null)
         {
-          if (!nseq.contains(map.getTo()))
+          if (map.getTo() != null && !nseq.contains(map.getTo()))
           {
             nseq.addElement(map.getTo());
           }
         }
       }
     }
-    if (nseq.size()>0) {
+    if (nseq.size() > 0)
+    {
       sequencesArray = new SequenceI[nseq.size()];
       nseq.toArray(sequencesArray);
     }
-- 
1.7.10.2