Test for a new version of the ecoTag sequence Matcher

src/ecoTag.py

@@ -92,8 +92,10 @@ def addSearchOptions(optionManager):
                              default=False,
                              help='Add an extra column in the output with the query sequence')
     
-    
-    
+    optionManager.add_option('--self-matches',
+                             action='store_true',dest='selfmatches',
+                             default=False,
+                             help='Switch to the new match algorithm')    
 
 
 def count(data):
@@ -135,6 +137,27 @@ def lcsIterator(entries,db,options):
         results = [x for x in results if x[1]>=minid]
         yield seq,maxid,results
 
+
+def lcsIteratorSelf(entries,db,options):
+    
+    for seq in entries:
+        results = []
+        maxid   = (None,0.0)
+        minid   = options.minimum
+        for d in db:
+            lcs,lali = lenlcs(seq,d,minid,normalized=True)
+            if lcs > maxid[1]:
+                maxid = (d,lcs)
+                minid = maxid[1] ** options.shape
+
+        for d in db:
+            if d.id != maxid[0].id:
+                lcs,lali = lenlcs(maxid[0],d,maxid[1],normalized=True)      
+                if lcs >= maxid[1]:
+                    results.append((d,lcs))
+                
+        yield seq,maxid,results
+        
 if __name__=='__main__':
     
     optionParser = getOptionManager([addSearchOptions,addTaxonomyDBOptions],
@@ -146,6 +169,7 @@ if __name__=='__main__':
     if options.explain is not None:
         options.table=True
         
+
     taxonomy = loadTaxonomyDatabase(options)
     
     print >>sys.stderr,"Reading reference DB ...",
@@ -177,7 +201,12 @@ if __name__=='__main__':
     if options.sort is not None:
         entries = sortSequence(entries, options.sort, options.reverse)
 
-    search = lcsIterator(entries,db,options)
+    matcher = lcsIterator
+    
+    if options.selfmatches:
+        matcher= lcsIteratorSelf
+
+    search = matcher(entries,db,options)
                      
                     
     for seq,best,match in search: