From b100e1ff9a2e736a5535dda3abded2bf59144af4 Mon Sep 17 00:00:00 2001
From: amwaterhouse <Andrew Waterhouse>
Date: Tue, 8 Feb 2005 18:40:35 +0000
Subject: [PATCH] Looks for missing ids using BlastClient

---
 src/jalview/analysis/SequenceFeatureFetcher.java |  101 +++++++++++++++++++++-
 1 file changed, 97 insertions(+), 4 deletions(-)

diff --git a/src/jalview/analysis/SequenceFeatureFetcher.java b/src/jalview/analysis/SequenceFeatureFetcher.java
index 8c550e9..e034b3f 100755
--- a/src/jalview/analysis/SequenceFeatureFetcher.java
+++ b/src/jalview/analysis/SequenceFeatureFetcher.java
@@ -10,9 +10,11 @@ public class SequenceFeatureFetcher implements Runnable
 {
   AlignmentI align;
   AlignmentPanel ap;
+  ArrayList unknownSequences;
 
   public SequenceFeatureFetcher(AlignmentI align, AlignmentPanel ap)
   {
+    unknownSequences = new ArrayList();
     this.align = align;
     this.ap = ap;
     Thread thread = new Thread(this);
@@ -78,7 +80,8 @@ public class SequenceFeatureFetcher implements Runnable
           remainingIds.append(ids.get(i) + ";");
 
          EBIFetchClient ebi = new EBIFetchClient();
-        String[] result = ebi.fetchData(remainingIds.toString(), "xml", null);
+         System.out.println(remainingIds.toString());
+         String[] result = ebi.fetchData(remainingIds.toString(), "xml", null);
 
         if(result!=null)
           ReadUniprotFile(result, out, align);
@@ -93,6 +96,13 @@ public class SequenceFeatureFetcher implements Runnable
     }
   }catch(Exception ex){ex.printStackTrace();}
 
+  ap.RefreshPanels();
+  findMissingIds(align);
+
+  if(unknownSequences.size()>0)
+  {
+    WSWUBlastClient blastClient = new WSWUBlastClient(align, unknownSequences);
+  }
 
 }
 
@@ -115,10 +125,21 @@ void ReadUniprotFile(String [] result, RandomAccessFile out, AlignmentI align)
         out.writeBytes("<entry>\n");
       }catch(Exception ex){}
 
-      sequence = align.findName( parseElement( result[r], "<name>" , out)) ;
+      String seqName = parseElement( result[r], "<name>" , out);
+      sequence = align.findName( seqName ) ;
       if(sequence==null)
       {
-        System.out.println("Couldnt find sequence id. Suggestion is "+result[r]);
+        sequence = align.findName( seqName.substring(0, seqName.indexOf('_')));
+        if(sequence!=null)
+        {
+          System.out.println("changing "+sequence.getName()+" to "+seqName);
+          sequence.setName(seqName);
+        }
+      }
+      if(sequence==null)
+      {
+        System.out.println("UNIPROT updated suggestion is "+result[r]);
+        sequence = align.findName( result[r] ) ;
 
         // this entry has been suggested by ebi.
         // doesn't match id in alignment file
@@ -180,6 +201,52 @@ void ReadUniprotFile(String [] result, RandomAccessFile out, AlignmentI align)
        }
      }
 
+     if(result[r].indexOf("<sequence")>-1)
+     {
+       StringBuffer seqString = new StringBuffer();
+
+       if(out!=null)
+         try  {  out.writeBytes(result[r]+"\n"); }   catch (Exception ex){}
+
+       while(result[++r].indexOf("</sequence>")==-1)
+       {
+         seqString.append(result[r]);
+         if(out!=null)
+         try  {  out.writeBytes(result[r]+"\n"); }   catch (Exception ex){}
+       }
+
+       if(out!=null)
+         try  {  out.writeBytes(result[r]+"\n"); }   catch (Exception ex){}
+
+       StringBuffer nonGapped = new StringBuffer();
+       for (int i = 0; i < sequence.getSequence().length(); i++)
+       {
+         if (!jalview.util.Comparison.isGap(sequence.getCharAt(i)))
+           nonGapped.append(sequence.getCharAt(i));
+       }
+
+       int absStart = seqString.toString().indexOf(nonGapped.toString());
+       if(absStart==-1)
+       {
+         unknownSequences.add(sequence.getName());
+         features = null;
+         System.out.println(sequence.getName()+ "does not match ");
+         continue;
+       }
+
+       int absEnd =   absStart + nonGapped.toString().length();
+       absStart+=1;
+
+       if(absStart!=sequence.getStart() || absEnd!=sequence.getEnd())
+         System.out.println("Updated: "+sequence.getName()+" "+
+                           sequence.getStart()+"/"+sequence.getEnd()+"  to  "+ absStart+"/"+absEnd);
+
+
+       sequence.setStart(absStart);
+       sequence.setEnd(absEnd);
+
+     }
+
      if(result[r].indexOf("</entry>")>-1)
      {
        if(features!=null)
@@ -191,8 +258,34 @@ void ReadUniprotFile(String [] result, RandomAccessFile out, AlignmentI align)
 
      }
   }
+}
+
+void findMissingIds(AlignmentI align)
+{
+  String data;
+  ArrayList cachedIds = new ArrayList();
+
+  try
+  {
+    BufferedReader in = new BufferedReader(
+        new FileReader(jalview.bin.Cache.getProperty("UNIPROT_CACHE")));
+
+    while ( (data = in.readLine()) != null)
+    {
+      if (data.indexOf("name") > -1)
+      {
+        String name = parseElement(data, "<name>", null);
+        cachedIds.add(name);
+      }
+    }
+  }
+  catch (Exception ex)
+  {   ex.printStackTrace();  }
+
+  for(int i=0; i<align.getHeight(); i++)
+    if( !cachedIds.contains( align.getSequenceAt(i).getName() ) )
+      unknownSequences.add( align.getSequenceAt(i).getName() );
 
-  ap.RefreshPanels();
 
 }
 
-- 
1.7.10.2