diff scripts/functions.py @ 7:f1e24200e5ae draft

planemo upload for repository htpps://github.com/abims-sbr/adaptearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c

--- a/scripts/functions.py	Wed Feb 28 06:06:14 2018 -0500
+++ b/scripts/functions.py	Wed Feb 28 10:39:41 2018 -0500
@@ -1,3 +1,5 @@
+import itertools
+
 def simplify_fasta_name(fasta_name,LT):
     for abbreviation in LT:
         if abbreviation in fasta_name:
@@ -7,35 +9,25 @@
 
 ## Generates bash, with key = fasta name; value = sequence (WITH GAP, IF ANY, REMOVED IN THIS FUNCTION)
 def dico(fasta_file,LT):
-
-    count_fastaName=0
-    F1 = open(fasta_file, "r")
-    
+    #count_fastaName = 0
     bash1 = {}
-    while 1:
-        nextline = F1.readline()
-        #print nextline
-        if not nextline :
-            break
-        
-        if nextline[0] == ">":
-            count_fastaName = count_fastaName + 1
-            fasta_name = nextline[1:-1]
-            nextline = F1.readline()
-            sequence = nextline[:-1]
-            
-            if fasta_name not in bash1.keys():
-                fasta_name = simplify_fasta_name(fasta_name,LT)  ### DEF 0 ###
-                bash1[fasta_name] = sequence
-            else:
-                print fasta_name
+    with open(fasta_file, "r") as file:
+        for name, query in itertools.izip_longest(*[file]*2):
+            if not name:
+                break
+            if name[0] == ">":
+                #count_fastaName += 1
+                fasta_name = name[1:-1]
+                sequence = query[:-1]
+                if fasta_name not in bash1.keys():
+                    fasta_name = simplify_fasta_name(fasta_name, LT)
+                    bash1[fasta_name] = sequence
+                else :
+                    print fasta_name
 
-    # Find alignment length
     kk = bash1.keys()
     key0 = kk[0]
     seq0 = bash1[key0]
     ln_seq = len(seq0)
-
-    F1.close()
     
-    return(bash1)
+    return(bash1)
\ No newline at end of file