From 2de36153a84ca246643cd28b2ce05e9f4159dd89 Mon Sep 17 00:00:00 2001
From: jprocter <Jim Procter>
Date: Mon, 4 Aug 2008 14:16:18 +0000
Subject: [PATCH] pfam seed and full alignment retrieval and nicer ordering of
 sequence db sources in combobox

---
 src/jalview/gui/SequenceFetcher.java            |    6 +--
 src/jalview/ws/SequenceFetcher.java             |   55 ++++++++++++++++++++++-
 src/jalview/ws/dbsources/Pfam.java              |   53 +++++++++++++---------
 src/jalview/ws/dbsources/PfamFull.java          |   43 ++++++++++++++++++
 src/jalview/ws/dbsources/PfamSeed.java          |   44 ++++++++++++++++++
 src/jalview/ws/seqfetcher/ASequenceFetcher.java |   20 +++++++++
 6 files changed, 195 insertions(+), 26 deletions(-)
 create mode 100644 src/jalview/ws/dbsources/PfamFull.java
 create mode 100644 src/jalview/ws/dbsources/PfamSeed.java

diff --git a/src/jalview/gui/SequenceFetcher.java b/src/jalview/gui/SequenceFetcher.java
index 53f75a4..a66ec2f 100755
--- a/src/jalview/gui/SequenceFetcher.java
+++ b/src/jalview/gui/SequenceFetcher.java
@@ -50,7 +50,7 @@ extends JPanel implements Runnable
   StringBuffer result;
   final String noDbSelected = "-- Select Database --";
   Hashtable sources = new Hashtable();
-  private static ASequenceFetcher sfetch=null;
+  private static jalview.ws.SequenceFetcher sfetch=null;
   public SequenceFetcher(IProgressIndicator guiIndic)
   {
       final IProgressIndicator guiWindow = guiIndic;
@@ -69,7 +69,7 @@ extends JPanel implements Runnable
             {
               guiWindow.setProgressBar("Initialising Sequence Database Fetchers", this.hashCode());
             }
-            ASequenceFetcher sf = new jalview.ws.SequenceFetcher();
+            jalview.ws.SequenceFetcher sf = new jalview.ws.SequenceFetcher();
             if ( guiWindow!=null )
             {
               guiWindow.setProgressBar("Initialising Sequence Database Fetchers", this.hashCode());
@@ -102,7 +102,7 @@ extends JPanel implements Runnable
      * and should be collapsed. 
      */
      
-    String dbs[] = sfetch.getSupportedDb();
+    String dbs[] = sfetch.getOrderedSupportedSources();
     for (int i=0; i<dbs.length;i++)
     {
       if (!sources.containsValue(dbs[i]))
diff --git a/src/jalview/ws/SequenceFetcher.java b/src/jalview/ws/SequenceFetcher.java
index 073d448..0d93cb8 100644
--- a/src/jalview/ws/SequenceFetcher.java
+++ b/src/jalview/ws/SequenceFetcher.java
@@ -1,5 +1,6 @@
 package jalview.ws;
 
+import java.util.ArrayList;
 import java.util.Enumeration;
 import java.util.Vector;
 
@@ -36,10 +37,62 @@ public class SequenceFetcher extends ASequenceFetcher
     addDBRefSourceImpl(jalview.ws.dbsources.Uniprot.class);
     addDBRefSourceImpl(jalview.ws.dbsources.UnprotName.class);
     addDBRefSourceImpl(jalview.ws.dbsources.Pdb.class);
-    addDBRefSourceImpl(jalview.ws.dbsources.Pfam.class);
+    addDBRefSourceImpl(jalview.ws.dbsources.PfamFull.class);
+    addDBRefSourceImpl(jalview.ws.dbsources.PfamSeed.class); // ensures Seed alignment is 'default' for PFAM
     registerDasSequenceSources();
   }
   /**
+   * return an ordered list of database sources suitable for using in a GUI element
+   */
+  public String[] getOrderedSupportedSources() {
+    String[] srcs = this.getSupportedDb();
+    ArrayList dassrc = new ArrayList(),
+      nondas=new ArrayList();
+    for (int i=0;i<srcs.length;i++)
+    {
+      String nm = getSourceProxy(srcs[i]).getDbName();
+      if (getSourceProxy(srcs[i]) instanceof jalview.ws.dbsources.DasSequenceSource)
+      {
+        if (nm.startsWith("das:"))
+        {
+          nm = nm.substring(4);
+        }
+        dassrc.add(new String[] { srcs[i], nm.toUpperCase()} );
+      } else {
+        nondas.add(new String[] { srcs[i], nm.toUpperCase()} );
+      }
+    }
+    Object[] sorted = nondas.toArray();
+    String[] tosort = new String[sorted.length];
+    nondas.clear();
+    for (int j=0;j<sorted.length;j++)
+    {
+      tosort[j] = ((String[]) sorted[j])[1];
+    }
+    jalview.util.QuickSort.sort(tosort, sorted);
+    int i=0;
+    for (int j=sorted.length-1;j>=0; j--,i++)
+    {
+      srcs[i] = ((String[]) sorted[j])[0];
+      sorted[j] = null;
+    }
+
+    sorted = dassrc.toArray();
+    tosort = new String[sorted.length];
+    dassrc.clear();
+    for (int j=0;j<sorted.length;j++)
+    {
+      tosort[j] = ((String[]) sorted[j])[1];
+    }
+    jalview.util.QuickSort.sort(tosort, sorted);
+    for (int j=sorted.length-1;j>=0; j--,i++)
+    {
+      srcs[i] = ((String[]) sorted[j])[0];
+      sorted[j] = null;
+    }
+    return srcs;
+  }
+  /**
    * simple run method to test dbsources.
    * @param argv
    */
diff --git a/src/jalview/ws/dbsources/Pfam.java b/src/jalview/ws/dbsources/Pfam.java
index db765eb..4136e62 100644
--- a/src/jalview/ws/dbsources/Pfam.java
+++ b/src/jalview/ws/dbsources/Pfam.java
@@ -9,23 +9,25 @@ import com.stevesoft.pat.Regex;
 
 import jalview.datamodel.AlignmentI;
 import jalview.datamodel.DBRefEntry;
+import jalview.io.AppletFormatAdapter;
 import jalview.io.FastaFile;
+import jalview.io.StockholmFile;
 import jalview.ws.seqfetcher.DbSourceProxy;
 import jalview.ws.seqfetcher.DbSourceProxyImpl;
 /**
- * TODO: later PFAM is a complex datasource - it currently returns a seed alignment, but could optionally return a full alignment.
  * TODO: later PFAM is a complex datasource - it could return a tree in addition to an alignment
- * TODO: HP: Incorporate jalview.gui.SequenceFetcher retrieval code here.
+ * TODO: create interface to pass alignment properties and tree back to sequence fetcher
  * @author JimP
  *
  */
-public class Pfam extends DbSourceProxyImpl implements DbSourceProxy
+abstract public class Pfam extends DbSourceProxyImpl implements DbSourceProxy
 {
 
   public Pfam()
   {
     super();
-    
+    // all extensions of this PFAM source base class are DOMAINDB sources 
+    addDbSourceProperty(jalview.datamodel.DBRefSource.DOMAINDB);
   }
 
   /* (non-Javadoc)
@@ -48,21 +50,24 @@ public class Pfam extends DbSourceProxyImpl implements DbSourceProxy
 
   /* (non-Javadoc)
    * @see jalview.ws.DbSourceProxy#getDbSource()
-   */
   public String getDbSource()
   {
+    ** this doesn't work - DbSource is key for the hash of DbSourceProxy instances - 1:many mapping for DbSource to proxy will be lost.
+    ** suggest : PFAM is an 'alignment' source - means proxy is higher level than a sequence source.
     return jalview.datamodel.DBRefSource.PFAM;
   }
+   */
 
+  
   /* (non-Javadoc)
    * @see jalview.ws.DbSourceProxy#getDbSourceProperties()
-   */
   public Hashtable getDbSourceProperties()
   {
-    // TODO Auto-generated method stub
+    
     return null;
   }
-
+ */
+  
   /* (non-Javadoc)
    * @see jalview.ws.DbSourceProxy#getDbVersion()
    */
@@ -71,7 +76,11 @@ public class Pfam extends DbSourceProxyImpl implements DbSourceProxy
     // TODO Auto-generated method stub
     return null;
   }
-  public static String PFAMURL = "http://www.sanger.ac.uk/cgi-bin/Pfam/getalignment.pl?format=fal&acc=";
+  /**
+   * 
+   * @return PFAM URL stub for this DbSource
+   */
+  protected abstract String getPFAMURL();
   /* (non-Javadoc)
    * @see jalview.ws.DbSourceProxy#getSequenceRecords(java.lang.String[])
    */
@@ -79,16 +88,20 @@ public class Pfam extends DbSourceProxyImpl implements DbSourceProxy
   {
     // TODO: this is not a perfect implementation. We need to be able to add individual references to each sequence in each family alignment that's retrieved. 
     startQuery();
-    results = new StringBuffer();
-    // split queries into many little ones.
-    results.append(new FastaFile(
-                PFAMURL+queries.trim().toUpperCase(), "URL").print());
-    stopQuery();
-    AlignmentI rcds = parseResult(results.toString());
+    AlignmentI rcds = new jalview.io.FormatAdapter().readFile(getPFAMURL()+queries.trim().toUpperCase(), jalview.io.FormatAdapter.URL,"STH");
     for (int s=0,sNum=rcds.getHeight(); s<sNum;s++)
     {
-      rcds.getSequenceAt(s).addDBRef(new DBRefEntry(getDbSource(), getDbVersion(), queries.trim().toUpperCase()));
+      rcds.getSequenceAt(s).addDBRef(new DBRefEntry(jalview.datamodel.DBRefSource.PFAM,
+              // getDbSource(), 
+              getDbVersion(), queries.trim().toUpperCase()));
+      if (!getDbSource().equals(jalview.datamodel.DBRefSource.PFAM))
+      {         // add the specific ref too
+        rcds.getSequenceAt(s).addDBRef(
+                new DBRefEntry( getDbSource(), 
+                        getDbVersion(), queries.trim().toUpperCase()));
+      }
     }
+    stopQuery();
     return rcds;
   }
 
@@ -100,13 +113,9 @@ public class Pfam extends DbSourceProxyImpl implements DbSourceProxy
     return accession.indexOf("PF")==0;
   }
 
-  public String getTestQuery()
-  {
-    return "PF00535";
-  }
 
-  public String getDbName()
+  /*public String getDbName()
   {
     return "PFAM"; // getDbSource();
-  }
+  } */
 }
diff --git a/src/jalview/ws/dbsources/PfamFull.java b/src/jalview/ws/dbsources/PfamFull.java
new file mode 100644
index 0000000..3d3ab53
--- /dev/null
+++ b/src/jalview/ws/dbsources/PfamFull.java
@@ -0,0 +1,43 @@
+/**
+ * 
+ */
+package jalview.ws.dbsources;
+
+import jalview.ws.seqfetcher.DbSourceProxy;
+
+/**
+ * flyweight class specifying retrieval of Full family alignments from PFAM
+ *
+ */
+public class PfamFull extends Pfam implements DbSourceProxy
+{
+  public PfamFull()
+  {
+    super();
+  }
+
+  /* (non-Javadoc)
+   * @see jalview.ws.dbsources.Pfam#getPFAMURL()
+   */
+  protected String getPFAMURL()
+  {
+    return "http://pfam.sanger.ac.uk/family/alignment/download/format?alnType=full&format=stockholm&order=t&case=l&gaps=default&entry=";
+  }
+
+  /* (non-Javadoc)
+   * @see jalview.ws.seqfetcher.DbSourceProxy#getDbName()
+   */
+  public String getDbName()
+  {
+    return "PFAM (Full)";
+  }
+  public String getDbSource()
+  {
+    return getDbName(); // so we have unique DbSource string.
+  }
+  public String getTestQuery()
+  {
+    return "PF00535"; // TODO: Pick a better full alignment to retrieve.
+  }
+
+}
diff --git a/src/jalview/ws/dbsources/PfamSeed.java b/src/jalview/ws/dbsources/PfamSeed.java
new file mode 100644
index 0000000..e0e2b4f
--- /dev/null
+++ b/src/jalview/ws/dbsources/PfamSeed.java
@@ -0,0 +1,44 @@
+/**
+ * 
+ */
+package jalview.ws.dbsources;
+
+import jalview.ws.seqfetcher.DbSourceProxy;
+
+/**
+ * flyweight class specifying retrieval of Seed alignments from PFAM
+ * @author JimP
+ *
+ */
+public class PfamSeed extends Pfam implements DbSourceProxy
+{
+  public PfamSeed()
+  {
+    super();
+  }
+
+  /* (non-Javadoc)
+   * @see jalview.ws.dbsources.Pfam#getPFAMURL()
+   */
+  protected String getPFAMURL()
+  {
+    return "http://pfam.sanger.ac.uk/family/alignment/download/format?alnType=seed&format=stockholm&order=t&case=l&gaps=default&entry=";
+  }
+
+  /* (non-Javadoc)
+   * @see jalview.ws.seqfetcher.DbSourceProxy#getDbName()
+   */
+  public String getDbName()
+  {
+    return "PFAM (Seed)";
+  }
+  public String getDbSource()
+  {
+    return jalview.datamodel.DBRefSource.PFAM; // archetype source
+  }
+  public String getTestQuery()
+  {
+    return "PF00535";
+  }
+
+}
diff --git a/src/jalview/ws/seqfetcher/ASequenceFetcher.java b/src/jalview/ws/seqfetcher/ASequenceFetcher.java
index c4f7e9a..0785ea2 100644
--- a/src/jalview/ws/seqfetcher/ASequenceFetcher.java
+++ b/src/jalview/ws/seqfetcher/ASequenceFetcher.java
@@ -255,4 +255,24 @@ public class ASequenceFetcher
     }
   }
 
+  /**
+   * test if the database handler for dbName contains the given dbProperty
+   * @param dbName
+   * @param dbProperty
+   * @return true if proxy has the given property
+   */
+  public boolean hasDbSourceProperty(String dbName, String dbProperty)
+  {
+    // TODO: decide if invalidDbName exception is thrown here.
+    DbSourceProxy proxy = getSourceProxy(dbName);
+    if (proxy!=null)
+    {
+      if (proxy.getDbSourceProperties()!=null)
+      {
+        return proxy.getDbSourceProperties().containsKey(dbProperty);
+      }
+    }
+    return false;
+  }
+
 }
\ No newline at end of file
-- 
1.7.10.2