|
0
|
1 #! /usr/bin/env python
|
|
|
2
|
|
|
3 #shuffleSequenceUsingAltschulErikson.py
|
|
|
4 #P. Clote, Oct 2003
|
|
|
5
|
|
|
6 #------------------------------------------------------------------
|
|
|
7 #Input RNAs in FASTA file, and compute NUM many shufflings of RNA sequence
|
|
|
8 #using Altman-Erikson randomly shuffled dinucleotide method.
|
|
|
9 #------------------------------------------------------------------
|
|
|
10
|
|
|
11 PRINT = 0
|
|
|
12 LINELEN = 70
|
|
|
13
|
|
|
14 import sys,os,stats,string
|
|
|
15 from altschulEriksonDinuclShuffle import dinuclShuffle
|
|
|
16 import computeRNAfoldEnergyForRNAsInFile
|
|
|
17
|
|
|
18
|
|
|
19
|
|
|
20
|
|
|
21 def file2string(fileName):
|
|
|
22 file = open(fileName,"r")
|
|
|
23 L = []
|
|
|
24 line = file.readline()
|
|
|
25 while line:
|
|
|
26 while line[0]==">": # treat lines beginning with '>' as comment and skip
|
|
|
27 line = file.readline()
|
|
|
28 continue
|
|
|
29 else:
|
|
|
30 line = line[:-1]
|
|
|
31 L.append(line)
|
|
|
32 line = file.readline()
|
|
|
33 text = string.join(L,"")
|
|
|
34 return text
|
|
|
35
|
|
|
36
|
|
|
37 def main(fileName,NUM):
|
|
|
38 seq = file2string(fileName)
|
|
|
39 for i in range(NUM):
|
|
|
40 shuffledSeq = dinuclShuffle(seq)
|
|
|
41 sys.stdout.write(">%d\n" % (i+1))
|
|
|
42 sys.stdout.write("%s\n" % shuffledSeq)
|
|
|
43
|
|
|
44
|
|
|
45
|
|
|
46
|
|
|
47 if __name__ == '__main__':
|
|
|
48 if len(sys.argv) < 3 :
|
|
|
49 print "Usage: %s RNAs.faa NUM" % sys.argv[0]
|
|
|
50 text = """
|
|
|
51 1) RNA.faa is FASTA file of ONE RNA sequence
|
|
|
52 2) NUM is number of random sequences to generate by
|
|
|
53 shuffling the dinucleotides of RNAs input
|
|
|
54 Script to compute Altman-Erikson randomly shuffled dinucleotides.
|
|
|
55 """
|
|
|
56 print text
|
|
|
57 sys.exit(1)
|
|
|
58 fileName = sys.argv[1]
|
|
|
59 NUM = int(sys.argv[2])
|
|
|
60 main(fileName,NUM)
|
|
|
61
|
