Changeset 16015

Timestamp:

05/30/12 10:36:57 (13 months ago)

Author:

coissac

Message:

 

Location:

trunk/src

Files:

• obitools/sample.py (modified) (2 diffs)
• obitools/utils/__init__.py (modified) (1 diff)
• obiuniq.py (modified) (1 diff)

trunk/src/obitools/sample.py

@@ -4 +4 @@
 @author: coissac
 '''
-from random import shuffle, randrange
+from random import shuffle, randrange, sample
+from obitools.utils import Counter
+import random
 
 def lookfor(x,cumsum):
@@ -53 +55 @@
 
 def weigthedSampleWithoutReplacement(events,size):
-    entries = [k for k in events.iterkeys() if events[k]>0]
-    shuffle(entries)
-    cumul=[]
+    # entries = [k for k in events.iterkeys() if events[k]>0]
+    entries = events.keys()
+    cumul=[0] * len(entries)
     s=0
+    i=0
     for e in entries:
         s+=events[e]
-        cumul.append(s)
+        cumul[i]=s
+        i+=1
     
-    #print cumul
-    result={}
+    c = random.sample(xrange(s),size)
+    c.sort()
     
-    for t in xrange(size):
-        # print s,cumul,
-        e=lookfor(randrange(s), cumul)
-        # print e
-        k=entries[e]
-        for x in xrange(e,len(cumul)):
-            cumul[x]-=1
-        s-=1
-        result[k]=result.get(k,0)+1
+    i = 0
+    for j in xrange(len(c)):
+        v = c[j]
+        while (v > cumul[i]):
+            i+=1
+        c[j]=entries[i]
+        
+    result=Counter(c)
 
     return result

trunk/src/obitools/utils/__init__.py

@@ -16 +16 @@
 from _utils import progressBar
 
+from collections import Counter
 
 

trunk/src/obiuniq.py

@@ -15 +15 @@
                              type="string",
                              default=[],
-                             help="ecoPCR taxonomy Database "
-                                  "name")
+                             help="attributes to merge")
     
     optionManager.add_option('-c','--category-attribute',