Add a --minimum-circle option to ecotag and a cache on the self

alignment scores of 1000000 of pairwise scores

setup.py

@@ -19,7 +19,7 @@ from os import path
 
 
 PACKAGE = "OBITools"
-VERSION = "1.1.18"
+VERSION = "1.1.19"
 AUTHOR  = 'Eric Coissac'
 EMAIL   = 'eric@coissac.eu'
 URL     = 'metabarcoding.org/obitools'

src/ecotag.py

@@ -46,6 +46,8 @@ from obitools.options.taxonomyfilter import addTaxonomyDBOptions,loadTaxonomyDat
 from obitools.options import getOptionManager
 from obitools.format.options import addInOutputOption, sequenceWriterGenerator
 
+from collections import OrderedDict
+
 import sys
 import math
 import os.path
@@ -76,6 +78,13 @@ def addSearchOptions(optionManager):
                              default=0.0,
                              help="minimum identity to consider.")
     
+    optionManager.add_option('--minimum-circle',
+                             action="store", dest="circle",
+                             metavar="identity",
+                             type="float",
+                             default=1.0,
+                             help="minimum identity considered for the assignment circle.")
+    
 #    optionManager.add_option('-S','--normalized-smallest',
 #                             action="store_false", dest="large",
 #                             default=True,
@@ -190,6 +199,26 @@ def myLenlcs(s1, s2, minid, normalized, reference):
     return lcs, lali
 
 
+def cachedLenLCS(s1,s2,minid,normalized,reference):
+    global __LCSCache__
+    
+    pair=frozenset((s1.id,s2.id))
+    
+    if pair in __LCSCache__:
+        rep=__LCSCache__[pair]
+        del __LCSCache__[pair]
+        
+    else:
+        rep=lenlcs(s1,s2,minid,normalized,reference)
+        
+    __LCSCache__[pair]=rep
+    
+    if len(__LCSCache__) > 1000000:
+        __LCSCache__.popitem(0)
+    return rep
+    
+
+
 #def lcsIterator(entries,db,options):
 #    
 #    for seq in entries:
@@ -233,7 +262,7 @@ def lcsIteratorSelf(entries,db,options):
         maxid   = ([],0.0)
         minid   = options.minimum
         for d in db:
-            lcs,lali = myLenlcs(seq,d,minid,normalized=True,reference=ALILEN)
+            lcs,lali = myLenlcs(seq,d,minid,normalized=True,reference=ALILEN)  # @UnusedVariable
             if lcs > maxid[1] and lcs > options.minimum:
                 maxid = ([d],lcs)
                 minid = maxid[1]
@@ -241,11 +270,13 @@ def lcsIteratorSelf(entries,db,options):
                 maxid[0].append(d)
                 
         if maxid[0]:
+            if maxid[1] > options.circle:
+                maxid[1]=options.circle
             results.extend([(s,maxid[1]) for s in maxid[0]])
             for d in db:
                 for s in maxid[0]:
                     if d.id != s.id:
-                        lcs,lali = lenlcs(s,d,maxid[1],normalized=True,reference=ALILEN)      
+                        lcs,lali = cachedLenLCS(s,d,maxid[1],normalized=True,reference=ALILEN)      # @UnusedVariable
                         if lcs >= maxid[1]:
                             results.append((d,lcs))
                 
@@ -253,6 +284,8 @@ def lcsIteratorSelf(entries,db,options):
         
 if __name__=='__main__':
     
+    __LCSCache__=OrderedDict()
+    
     optionParser = getOptionManager([addSearchOptions,addTaxonomyDBOptions,addInOutputOption],progdoc=__doc__)
     
     (options, entries) = optionParser()

src/obitools/version.py

@@ -1,5 +1,5 @@
 major = 1
 minor = 1
-serial= '18'
+serial= '19'
 
 version = "%2d.%02d %s" % (major,minor,serial)