h

baptiste.py

@@ -4,12 +4,31 @@ import csv
 from Bio.Seq import Seq
 from Bio.Alphabet import generic_dna
 import sys
+from collections import Counter
+
+def cond(hyb,start,stop):
+    s = hyb[:180]
+    i = ind(s,stop)
+    if i>=0:
+        sub_s = s[i:]
+        j = ind(sub_s,start)
+        if j<0:
+            return False
+        elif 180-j<30:
+            return False
+        elif 180-j>=30:
+            return True
+    return False
+
+
 
 if __name__ == "__main__":
 	reader = csv.reader(open("./Pseudomonas_aeruginosa_UCBPP-PA14_109.csv","r"))
 	web_gene = list(open("./Pseudomonas_aeruginosa_UCBPP-PA14_109.fna","r"))[1:]
 	web_gene = "".join([e.strip() for e in web_gene])
 	data = list(reader)
+	web_ba_seqs = list()
+	hybrids = list()
 	for e in data[3:]:
 		if len(e)>=5:
 			_from = int(e[3]) - 1
@@ -18,4 +37,9 @@ if __name__ == "__main__":
 			_ba = e[-2]
 			_aa = e[-1]
 			seq = Seq(_ba,generic_dna)
+			web_ba_seqs.append(seq)
 			hybrid = Seq(web_gene[_from-180:_to],generic_dna)
+			hybrids.append(hybrid)
+	start_codons = Counter([seq[:3] for seq in web_ba_seqs])
+	stop_codons = Counter([seq[-3:] for seq in web_ba_seqs])
+