comparison Iterative_mapping/iterative_map.py @ 56:9d26c2e4953e draft

Uploaded
author tyty
date Tue, 18 Nov 2014 01:00:33 -0500
parents
children
comparison
equal deleted inserted replaced
55:20b8f6ab0e05 56:9d26c2e4953e
1 #!/usr/bin/env python
2 # -*- coding: utf-8 -*-
3
4 import sys
5 import os
6 from read_file import *
7 from read_s_file import *
8 import random
9 import string
10
11 type_input = sys.argv[1]
12 seq_file = sys.argv[2]
13 ref_file = sys.argv[3]
14 shift = sys.argv[4]
15 length = sys.argv[5]
16 t_end = sys.argv[6]
17 map_type = sys.argv[7]
18 output_file = sys.argv[8]
19
20
21 if map_type!="default":
22 s = ""
23 s = s+"-v "+sys.argv[9]
24 s = s+" -5 "+sys.argv[10]
25 s = s+" -3 "+sys.argv[11]
26 s = s+" -k "+sys.argv[12]
27 if sys.argv[13]:
28 s = s+" -a"
29 if int(sys.argv[14])>=1:
30 s = s+" -m "+sys.argv[14]
31 if sys.argv[15]:
32 s = s+" --best --strata "
33
34 else:
35 s = "-v 3 -a --best --strata "
36
37 ospath = os.path.realpath(sys.argv[0])
38 ost = ospath.split('/')
39 syspath = ""
40 for i in range(len(ost)-1):
41 syspath = syspath+ost[i].strip()
42 syspath = syspath+'/'
43
44 rs = ''.join(random.sample(string.ascii_letters + string.digits, 8))
45 os.system("mkdir "+syspath+"output_"+rs)
46
47 syspathrs = syspath+"output_"+rs+'/'
48
49 os.system("bowtie-build -f "+ref_file+" "+syspathrs+"ref > "+syspathrs+"log.txt")
50
51 os.system("cp "+seq_file+" "+syspathrs+"seq0.fa")
52
53 if type_input == "fasta":
54 tp = 'fasta'
55 if type_input == "fastq":
56 tp = 'fastq'
57
58 k = 0
59 print(type_input)
60 while(True):
61 if type_input == "fasta":
62 os.system("bowtie "+s+"-f "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam")
63 if type_input == "fastq":
64 os.system("bowtie "+s+"-q "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam")
65 os.system("samtools view -Sb -F 0xfff "+syspathrs+"map"+str(k)+".sam > "+syspathrs+"mapped"+str(k)+".bam 2>"+syspathrs+"log.txt") #get mapped reads
66 os.system("samtools view -Sb -f 0x4 "+syspathrs+"map"+str(k)+".sam > "+syspathrs+"umapped"+str(k)+".bam 2>"+syspathrs+"log.txt") #get unmapped reads
67 os.system("samtools view -Sb -f 0x10 "+syspathrs+"map"+str(k)+".sam > "+syspathrs+"rmapped"+str(k)+".bam 2>"+syspathrs+"log.txt") #get reversed mapped reads
68 os.system("samtools merge -f "+syspathrs+"unmapped"+str(k)+".bam "+syspathrs+"umapped"+str(k)+".bam "+syspathrs+"rmapped"+str(k)+".bam") #get reversed mapped reads
69 os.system("samtools view -h -o "+syspathrs+"unmapped"+str(k)+".sam "+syspathrs+"unmapped"+str(k)+".bam") #get reversed mapped reads
70 if k>0:
71 os.system("samtools view -h -o "+syspathrs+"mapped"+str(k)+".sam "+syspathrs+"mapped"+str(k)+".bam") #get reversed mapped reads
72 os.system("cut -f 1 "+syspathrs+"unmapped"+str(k)+".sam > "+syspathrs+"unmapped"+str(k)+".txt")
73 os.system("cut -f 1 "+syspathrs+"mapped"+str(k)+".sam > "+syspathrs+"mapped"+str(k)+".txt")
74 os.system("python "+syspath+"remove_map.py "+syspathrs+"unmapped"+str(k)+".txt "+syspathrs+"mapped"+str(k)+".txt "+syspathrs+"runmapped"+str(k)+".txt")
75 os.system("rm "+syspathrs+"mapped"+str(k)+".sam")
76 os.system("rm "+syspathrs+"mapped"+str(k)+".txt")
77 os.system("rm "+syspathrs+"unmapped"+str(k)+".txt")
78 else:
79 os.system("cut -f 1 "+syspathrs+"unmapped"+str(k)+".sam > "+syspathrs+"runmapped"+str(k)+".txt")
80
81 os.system("rm "+syspathrs+"unmapped"+str(k)+".bam")
82 os.system("rm "+syspathrs+"umapped"+str(k)+".bam")
83 os.system("rm "+syspathrs+"rmapped"+str(k)+".bam")
84 os.system("python "+syspath+"seq_track.py "+syspathrs+"runmapped"+str(k)+".txt "+syspathrs+"seq"+str(k)+".fa "+syspathrs+"unmap_seq"+str(k)+".fa "+tp) #get unmapped sequence
85 os.system("python "+syspath+"truncate.py "+syspathrs+"unmap_seq"+str(k)+".fa "+shift+" "+syspathrs+"seq"+str(k+1)+".fa "+length+" "+t_end) #truncate unmapped sequence
86 os.system("rm "+syspathrs+"seq"+str(k)+".fa") #Remove sequences being mapped
87 os.system("rm "+syspathrs+"map"+str(k)+".sam") #Remove mapping file
88 os.system("rm "+syspathrs+"unmap_seq"+str(k)+".fa") #Remove unmapped sequnce
89 os.system("rm "+syspathrs+"runmapped"+str(k)+".txt")
90 os.system("rm "+syspathrs+"unmapped"+str(k)+".sam")
91
92 os.system("wc -l "+syspathrs+"seq"+str(k+1)+".fa > "+syspathrs+"count"+str(k+1)+".txt")
93 c = read_sp_file(syspathrs+"count"+str(k+1)+".txt")
94 if c[0][0] == '0': #If no reads is in the sequence file, stop
95 os.system("rm "+syspathrs+"count"+str(k+1)+".txt")
96 os.system("rm "+syspathrs+"seq"+str(k+1)+".fa")
97 break
98 os.system("rm "+syspathrs+"count"+str(k+1)+".txt")
99 k = k+1
100
101 ss = ""
102 for i in range(0,k+1):
103 ss = ss+" "+syspathrs+"mapped"+str(i)+".bam"
104
105
106 os.system("samtools merge -f "+output_file+" "+ss)
107 #print("samtools merge mapped_all.bam"+ss)
108 os.system("rm "+syspathrs+"mapped*.bam")
109 os.system("rm "+syspathrs+"ref*")
110 os.system("rm -r "+syspathrs)
111
112