From 006226fa282af3eed4c9770a20073b0f4fa990d4 Mon Sep 17 00:00:00 2001
From: jprocter <Jim Procter>
Date: Fri, 7 Sep 2007 16:58:45 +0000
Subject: [PATCH] formatting and ensure that dataset is actually searched for
 sequences with dbrefs that correspond to a dbref
 crossreference sequence being retrieved (still outstanding
 bug for embl->proteinprod->emblcds retrieval

---
 src/jalview/analysis/CrossRef.java |  259 +++++++++++++++++++-----------------
 1 file changed, 138 insertions(+), 121 deletions(-)

diff --git a/src/jalview/analysis/CrossRef.java b/src/jalview/analysis/CrossRef.java
index f475ecb..d2f0358 100644
--- a/src/jalview/analysis/CrossRef.java
+++ b/src/jalview/analysis/CrossRef.java
@@ -39,27 +39,29 @@ public class CrossRef
     else
     {
       rfs = jalview.util.DBRefUtils.selectRefs(rfs,
-              DBRefSource.DNACODINGDBS); // could attempt to find other cross refs and return here - ie PDB xrefs (not dna, not protein seq)
+              DBRefSource.DNACODINGDBS); // could attempt to find other cross
+                                          // refs and return here - ie PDB xrefs
+                                          // (not dna, not protein seq)
     }
     return rfs;
   }
 
-
   public static Hashtable classifyDbRefs(DBRefEntry[] rfs)
   {
     Hashtable classes = new Hashtable();
-    classes.put(DBRefSource.PROTEINDBS, jalview.util.DBRefUtils.selectRefs(rfs, DBRefSource.PROTEINDBS));
-    classes.put(DBRefSource.DNACODINGDBS, jalview.util.DBRefUtils.selectRefs(rfs,
-            DBRefSource.DNACODINGDBS));
-    classes.put(DBRefSource.DOMAINDBS, jalview.util.DBRefUtils.selectRefs(rfs,
-            DBRefSource.DOMAINDBS));
+    classes.put(DBRefSource.PROTEINDBS, jalview.util.DBRefUtils.selectRefs(
+            rfs, DBRefSource.PROTEINDBS));
+    classes.put(DBRefSource.DNACODINGDBS, jalview.util.DBRefUtils
+            .selectRefs(rfs, DBRefSource.DNACODINGDBS));
+    classes.put(DBRefSource.DOMAINDBS, jalview.util.DBRefUtils.selectRefs(
+            rfs, DBRefSource.DOMAINDBS));
     // classes.put(OTHER, )
     return classes;
   }
 
   /**
    * @param dna
-   *          true if seqs are DNA seqs
+   *                true if seqs are DNA seqs
    * @param seqs
    * @return a list of sequence database cross reference source types
    */
@@ -67,22 +69,25 @@ public class CrossRef
   {
     return findSequenceXrefTypes(dna, seqs, null);
   }
+
   /**
-   * Indirect references are references from other sequences from the dataset to any of the direct
-   * DBRefEntrys on the given sequences.
+   * Indirect references are references from other sequences from the dataset to
+   * any of the direct DBRefEntrys on the given sequences.
+   * 
    * @param dna
-   *          true if seqs are DNA seqs
+   *                true if seqs are DNA seqs
    * @param seqs
    * @return a list of sequence database cross reference source types
    */
-  public static String[] findSequenceXrefTypes(boolean dna, SequenceI[] seqs, AlignmentI dataset)
+  public static String[] findSequenceXrefTypes(boolean dna,
+          SequenceI[] seqs, AlignmentI dataset)
   {
     String[] dbrefs = null;
     Vector refs = new Vector();
     for (int s = 0; s < seqs.length; s++)
     {
       SequenceI dss = seqs[s];
-      while (dss.getDatasetSequence()!=null)
+      while (dss.getDatasetSequence() != null)
       {
         dss = dss.getDatasetSequence();
       }
@@ -94,18 +99,19 @@ public class CrossRef
           refs.addElement(rfs[r].getSource());
         }
       }
-      if (dataset!=null)
+      if (dataset != null)
       {
         // search for references to this sequence's direct references.
         DBRefEntry[] lrfs = CrossRef.findXDbRefs(!dna, seqs[s].getDBRef());
         Vector rseqs = new Vector();
-        CrossRef.searchDatasetXrefs(seqs[s], !dna, lrfs, dataset, rseqs, null); // don't need to specify codon frame for mapping here
+        CrossRef.searchDatasetXrefs(seqs[s], !dna, lrfs, dataset, rseqs,
+                null); // don't need to specify codon frame for mapping here
         Enumeration lr = rseqs.elements();
         while (lr.hasMoreElements())
         {
           SequenceI rs = (SequenceI) lr.nextElement();
           DBRefEntry[] xrs = findXDbRefs(dna, rs.getDBRef());
-          for (int r=0; rfs != null && r < rfs.length; r++)
+          for (int r = 0; rfs != null && r < rfs.length; r++)
           {
             if (!refs.contains(rfs[r].getSource()))
             {
@@ -188,7 +194,7 @@ public class CrossRef
    * @param dna
    * @param source
    * @param dataset
-   *          alignment to search for product sequences.
+   *                alignment to search for product sequences.
    * @return products (as dataset sequences)
    */
   public static Alignment findXrefSequences(SequenceI[] seqs, boolean dna,
@@ -196,24 +202,32 @@ public class CrossRef
   {
     Vector rseqs = new Vector();
     Alignment ral = null;
-    AlignedCodonFrame cf=new AlignedCodonFrame(0); // nominal width
+    AlignedCodonFrame cf = new AlignedCodonFrame(0); // nominal width
     for (int s = 0; s < seqs.length; s++)
     {
       SequenceI dss = seqs[s];
-      while (dss.getDatasetSequence()!=null)
+      while (dss.getDatasetSequence() != null)
       {
         dss = dss.getDatasetSequence();
       }
       boolean found = false;
       DBRefEntry[] xrfs = CrossRef.findXDbRefs(dna, dss.getDBRef());
-      if ((xrfs == null || xrfs.length == 0) && dataset!=null)
+      if ((xrfs == null || xrfs.length == 0) && dataset != null)
       {
         System.out.println("Attempting to find ds Xrefs refs.");
-        DBRefEntry[] lrfs = CrossRef.findXDbRefs(!dna, seqs[s].getDBRef()); // less ambiguous would be a 'find primary dbRefEntry' method.
+        DBRefEntry[] lrfs = CrossRef.findXDbRefs(!dna, seqs[s].getDBRef()); // less
+                                                                            // ambiguous
+                                                                            // would
+                                                                            // be a
+                                                                            // 'find
+                                                                            // primary
+                                                                            // dbRefEntry'
+                                                                            // method.
         // filter for desired source xref here
-        found = CrossRef.searchDatasetXrefs(dss, !dna, lrfs, dataset, rseqs, cf);
+        found = CrossRef.searchDatasetXrefs(dss, !dna, lrfs, dataset,
+                rseqs, cf);
       }
-      for (int r = 0; xrfs!=null && r < xrfs.length; r++)
+      for (int r = 0; xrfs != null && r < xrfs.length; r++)
       {
         if (source != null && !source.equals(xrfs[r].getSource()))
           continue;
@@ -223,14 +237,17 @@ public class CrossRef
           {
             Sequence rsq = new Sequence(xrfs[r].getMap().getTo());
             rseqs.addElement(rsq);
-            if (xrfs[r].getMap().getMap().getFromRatio()!=xrfs[r].getMap().getMap().getToRatio())
+            if (xrfs[r].getMap().getMap().getFromRatio() != xrfs[r]
+                    .getMap().getMap().getToRatio())
             {
               // get sense of map correct for adding to product alignment.
               if (dna)
               {
                 // map is from dna seq to a protein product
                 cf.addMap(dss, rsq, xrfs[r].getMap().getMap());
-              } else {
+              }
+              else
+              {
                 // map should be from protein seq to its coding dna
                 cf.addMap(rsq, dss, xrfs[r].getMap().getMap().getInverse());
               }
@@ -238,13 +255,13 @@ public class CrossRef
             found = true;
           }
         }
-        else
+        if (!found)
         {
           // do a bit more work - search for sequences with references matching
           // xrefs on this sequence.
           if (dataset != null)
           {
-            found = searchDataset(dss, xrfs[r], dataset, rseqs, cf);
+            found |= searchDataset(dss, xrfs[r], dataset, rseqs, cf);
             if (found)
               xrfs[r] = null; // we've recovered seqs for this one.
           }
@@ -265,8 +282,10 @@ public class CrossRef
           for (int r = 0; r < xrfs.length; r++)
           {
             // filter out any irrelevant or irretrievable references
-            if (xrfs[r]==null || ((source != null && !source.equals(xrfs[r].getSource()))
-                    || !sftch.isFetchable(xrfs[r].getSource())))
+            if (xrfs[r] == null
+                    || ((source != null && !source.equals(xrfs[r]
+                            .getSource())) || !sftch.isFetchable(xrfs[r]
+                            .getSource())))
             {
               l--;
               xrfs[r] = null;
@@ -275,7 +294,7 @@ public class CrossRef
           if (l > 0)
           {
             System.out
-            .println("Attempting to retrieve cross referenced sequences.");
+                    .println("Attempting to retrieve cross referenced sequences.");
             DBRefEntry[] t = new DBRefEntry[l];
             l = 0;
             for (int r = 0; r < xrfs.length; r++)
@@ -290,8 +309,8 @@ public class CrossRef
             } catch (Exception e)
             {
               System.err
-              .println("Problem whilst retrieving cross references for Sequence : "
-                      + seqs[s].getName());
+                      .println("Problem whilst retrieving cross references for Sequence : "
+                              + seqs[s].getName());
               e.printStackTrace();
             }
             if (retrieved != null)
@@ -310,7 +329,7 @@ public class CrossRef
       SequenceI[] rsqs = new SequenceI[rseqs.size()];
       rseqs.copyInto(rsqs);
       ral = new Alignment(rsqs);
-      if (cf!=null && cf.getProtMappings()!=null)
+      if (cf != null && cf.getProtMappings() != null)
       {
         ral.addCodonFrame(cf);
       }
@@ -319,20 +338,24 @@ public class CrossRef
   }
 
   /**
-   * find references to lrfs in the cross-reference set of each sequence in dataset (that is not equal to sequenceI)
-   * Identifies matching DBRefEntry based on source and accession string only - Map and Version are nulled.
+   * find references to lrfs in the cross-reference set of each sequence in
+   * dataset (that is not equal to sequenceI) Identifies matching DBRefEntry
+   * based on source and accession string only - Map and Version are nulled.
+   * 
    * @param sequenceI
    * @param lrfs
    * @param dataset
    * @param rseqs
    * @return true if matches were found.
    */
-  private static boolean searchDatasetXrefs(SequenceI sequenceI, boolean dna, DBRefEntry[] lrfs, AlignmentI dataset, Vector rseqs, AlignedCodonFrame cf)
+  private static boolean searchDatasetXrefs(SequenceI sequenceI,
+          boolean dna, DBRefEntry[] lrfs, AlignmentI dataset, Vector rseqs,
+          AlignedCodonFrame cf)
   {
-    boolean found=false;
-    if (lrfs==null)
+    boolean found = false;
+    if (lrfs == null)
       return false;
-    for (int i=0;i<lrfs.length; i++)
+    for (int i = 0; i < lrfs.length; i++)
     {
       DBRefEntry xref = new DBRefEntry(lrfs[i]);
       // add in wildcards
@@ -343,7 +366,6 @@ public class CrossRef
     return found;
   }
 
-
   /**
    * search a given sequence dataset for references matching cross-references to
    * the given sequence
@@ -352,34 +374,39 @@ public class CrossRef
    * @param xrf
    * @param dataset
    * @param rseqs
+   *                set of unique sequences
    * @param cf
-   * @return true if sequences were found and added
+   * @return true if one or more unique sequences were found and added
    */
   public static boolean searchDataset(SequenceI sequenceI, DBRefEntry xrf,
           AlignmentI dataset, Vector rseqs, AlignedCodonFrame cf)
   {
-    return searchDataset(sequenceI, xrf,
-            dataset, rseqs, cf, true, false);
+    return searchDataset(sequenceI, xrf, dataset, rseqs, cf, true, false);
   }
+
   /**
-   * TODO: generalise to different protein classifications
-   * Search dataset for DBRefEntrys matching the given one (xrf) and add
-   * the associated sequence to rseq.
+   * TODO: generalise to different protein classifications Search dataset for
+   * DBRefEntrys matching the given one (xrf) and add the associated sequence to
+   * rseq.
+   * 
    * @param sequenceI
    * @param xrf
    * @param dataset
    * @param rseqs
-   * @param direct - search all references or only subset
-   * @param dna search dna or protein xrefs (if direct=false)
+   * @param direct -
+   *                search all references or only subset
+   * @param dna
+   *                search dna or protein xrefs (if direct=false)
    * @return true if relationship found and sequence added.
    */
   public static boolean searchDataset(SequenceI sequenceI, DBRefEntry xrf,
-          AlignmentI dataset, Vector rseqs, AlignedCodonFrame cf, boolean direct, boolean dna)
+          AlignmentI dataset, Vector rseqs, AlignedCodonFrame cf,
+          boolean direct, boolean dna)
   {
     boolean found = false;
-    if (dataset==null) 
+    if (dataset == null)
       return false;
-    if (dataset.getSequences()==null)
+    if (dataset.getSequences() == null)
     {
       System.err.println("Empty dataset sequence set - NO VECTOR");
       return false;
@@ -393,46 +420,58 @@ public class CrossRef
         if (nxt.getDatasetSequence() != null)
         {
           System.err
-          .println("Implementation warning: getProducts passed a dataset alignment without dataset sequences in it!");
+                  .println("Implementation warning: getProducts passed a dataset alignment without dataset sequences in it!");
         }
         if (nxt != sequenceI && nxt != sequenceI.getDatasetSequence())
         {
-          DBRefEntry[] poss=null, cands=null;
+          // look for direct or indirect references in common
+          DBRefEntry[] poss = null, cands = null;
           if (direct)
           {
-            cands = jalview.util.DBRefUtils.searchRefs(poss=nxt
+            cands = jalview.util.DBRefUtils.searchRefs(poss = nxt
                     .getDBRef(), xrf);
-          } else {
-            cands = jalview.util.DBRefUtils.searchRefs(
-                    poss=CrossRef.findXDbRefs(dna, nxt.getDBRef()), xrf);
+          }
+          else
+          {
+            cands = jalview.util.DBRefUtils.searchRefs(poss = CrossRef
+                    .findXDbRefs(dna, nxt.getDBRef()), xrf);
           }
           if (cands != null)
           {
-            rseqs.addElement(nxt);
-            boolean foundmap= cf!=null; // don't search if we aren't given a codon map object
-            for (int r=0; foundmap && r<cands.length; r++)
+            if (!rseqs.contains(nxt))
             {
-              if (cands[r].hasMap())
+              rseqs.addElement(nxt);
+              boolean foundmap = cf != null; // don't search if we aren't given
+                                              // a codon map object
+              for (int r = 0; foundmap && r < cands.length; r++)
               {
-                if (cands[r].getMap().getTo()!=null && cands[r].getMap().getMap().getFromRatio()!=cands[r].getMap().getMap().getToRatio())
+                if (cands[r].hasMap())
                 {
-                  foundmap=true;
-                  // get sense of map correct for adding to product alignment.
-                  if (dna)
+                  if (cands[r].getMap().getTo() != null
+                          && cands[r].getMap().getMap().getFromRatio() != cands[r]
+                                  .getMap().getMap().getToRatio())
                   {
-                    // map is from dna seq to a protein product
-                    cf.addMap(sequenceI, nxt, cands[r].getMap().getMap()); 
-                  } else {
-                    // map should be from protein seq to its coding dna
-                    cf.addMap(nxt, sequenceI, cands[r].getMap().getMap().getInverse());
+                    foundmap = true;
+                    // get sense of map correct for adding to product alignment.
+                    if (dna)
+                    {
+                      // map is from dna seq to a protein product
+                      cf.addMap(sequenceI, nxt, cands[r].getMap().getMap());
+                    }
+                    else
+                    {
+                      // map should be from protein seq to its coding dna
+                      cf.addMap(nxt, sequenceI, cands[r].getMap().getMap()
+                              .getInverse());
+                    }
                   }
                 }
               }
+              // TODO: add mapping between sequences if necessary
+              found = true;
             }
-            // TODO: add mapping between sequences if necessary
-            found = true;
           }
-          
+
         }
       }
     }
@@ -440,56 +479,34 @@ public class CrossRef
   }
 
   /**
-   * precalculate different products that can be found for seqs in dataset
-   * and return them.
+   * precalculate different products that can be found for seqs in dataset and
+   * return them.
+   * 
    * @param dna
    * @param seqs
    * @param dataset
-   * @param fake - don't actually build lists - just get types
-   * @return
-  public static Object[] buildXProductsList(boolean dna, SequenceI[] seqs, AlignmentI dataset, boolean fake)
-  {
-    String types[] = jalview.analysis.CrossRef.findSequenceXrefTypes(
-            dna, seqs, dataset);
-    if (types != null)
-    {
-      System.out.println("Xref Types for: "+(dna ? "dna" : "prot"));
-      for (int t = 0; t < types.length; t++)
-      {
-        System.out.println("Type: " + types[t]);
-        SequenceI[] prod = 
-          jalview.analysis.CrossRef.findXrefSequences(seqs, dna, types[t]);
-        System.out.println("Found "
-                + ((prod == null) ? "no" : "" + prod.length)
-                + " products");
-        if (prod!=null)
-        {
-          for (int p=0; p<prod.length; p++)
-          {
-            System.out.println("Prod "+p+": "+prod[p].getDisplayId(true));
-          }
-        }
-      }
-
-    } else {
-      System.out.println("Trying getProducts for "+al.getSequenceAt(0).getDisplayId(true));
-      System.out.println("Search DS Xref for: "+(dna ? "dna" : "prot"));
-      // have a bash at finding the products amongst all the retrieved sequences.
-      SequenceI[] prod = jalview.analysis.CrossRef.findXrefSequences(al
-              .getSequencesArray(), dna, null, ds);
-      System.out.println("Found "
-              + ((prod == null) ? "no" : "" + prod.length)
-              + " products");
-      if (prod!=null)
-      {
-        // select non-equivalent sequences from dataset list
-        for (int p=0; p<prod.length; p++)
-        {
-          System.out.println("Prod "+p+": "+prod[p].getDisplayId(true));
-        }
-      }
-
-    }
-  }
+   * @param fake -
+   *                don't actually build lists - just get types
+   * @return public static Object[] buildXProductsList(boolean dna, SequenceI[]
+   *         seqs, AlignmentI dataset, boolean fake) { String types[] =
+   *         jalview.analysis.CrossRef.findSequenceXrefTypes( dna, seqs,
+   *         dataset); if (types != null) { System.out.println("Xref Types for:
+   *         "+(dna ? "dna" : "prot")); for (int t = 0; t < types.length; t++) {
+   *         System.out.println("Type: " + types[t]); SequenceI[] prod =
+   *         jalview.analysis.CrossRef.findXrefSequences(seqs, dna, types[t]);
+   *         System.out.println("Found " + ((prod == null) ? "no" : "" +
+   *         prod.length) + " products"); if (prod!=null) { for (int p=0; p<prod.length;
+   *         p++) { System.out.println("Prod "+p+":
+   *         "+prod[p].getDisplayId(true)); } } }
+   *  } else { System.out.println("Trying getProducts for
+   * "+al.getSequenceAt(0).getDisplayId(true)); System.out.println("Search DS
+   * Xref for: "+(dna ? "dna" : "prot")); // have a bash at finding the products
+   * amongst all the retrieved sequences. SequenceI[] prod =
+   * jalview.analysis.CrossRef.findXrefSequences(al .getSequencesArray(), dna,
+   * null, ds); System.out.println("Found " + ((prod == null) ? "no" : "" +
+   * prod.length) + " products"); if (prod!=null) { // select non-equivalent
+   * sequences from dataset list for (int p=0; p<prod.length; p++) {
+   * System.out.println("Prod "+p+": "+prod[p].getDisplayId(true)); } }
+   *  } }
    */
 }
\ No newline at end of file
-- 
1.7.10.2