comparison Iterative_mapping/iterative_map.py @ 66:d2817a631a7b draft

Uploaded
author tyty
date Tue, 18 Nov 2014 16:24:04 -0500
parents
children
comparison
equal deleted inserted replaced
65:36d912d5b1ac 66:d2817a631a7b
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 sm = ""
24 s = s+"-v "+sys.argv[9]
25 sm = sm+"-v "+sys.argv[9]
26 sm = sm+" -5 "+sys.argv[10]
27 sm = sm+" -3 "+sys.argv[11]
28 s = s+" -k "+sys.argv[12]
29 sm = sm+" -k "+sys.argv[12]
30 if sys.argv[13]:
31 s = s+" -a"
32 sm = sm+" -a"
33 if int(sys.argv[14])>=1:
34 s = s+" -m "+sys.argv[14]
35 sm = sm+" -m "+sys.argv[14]
36 if sys.argv[15]:
37 s = s+" --best --strata "
38 sm = sm+" --best --strata "
39
40 else:
41 s = "-v 3 -a --best --strata "
42 sm = "-v 3 -a --best --strata "
43
44 ospath = os.path.realpath(sys.argv[0])
45 ost = ospath.split('/')
46 syspath = ""
47 for i in range(len(ost)-1):
48 syspath = syspath+ost[i].strip()
49 syspath = syspath+'/'
50
51 syspathrs = os.getcwd()
52
53 os.system("bowtie-build -f "+ref_file+" "+syspathrs+"ref > "+syspathrs+"log.txt")
54
55 os.system("cp "+seq_file+" "+syspathrs+"seq0.fa")
56
57 if type_input == "fasta":
58 tp = 'fasta'
59 if type_input == "fastq":
60 tp = 'fastq'
61
62 k = 0
63
64 if type_input == "fasta":
65 os.system("bowtie "+sm+"-f "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam")
66 if type_input == "fastq":
67 os.system("bowtie "+sm+"-q "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam")
68
69 while(True):
70 os.system("samtools view -Sb -F 0xfff "+syspathrs+"map"+str(k)+".sam > "+syspathrs+"mapped"+str(k)+".bam 2>"+syspathrs+"log.txt") #get mapped reads
71 os.system("samtools view -Sb -f 0x4 "+syspathrs+"map"+str(k)+".sam > "+syspathrs+"umapped"+str(k)+".bam 2>"+syspathrs+"log.txt") #get unmapped reads
72 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
73 os.system("samtools merge -f "+syspathrs+"unmapped"+str(k)+".bam "+syspathrs+"umapped"+str(k)+".bam "+syspathrs+"rmapped"+str(k)+".bam") #get reversed mapped reads
74 os.system("samtools view -h -o "+syspathrs+"unmapped"+str(k)+".sam "+syspathrs+"unmapped"+str(k)+".bam") #get reversed mapped reads
75 if k>0:
76 os.system("samtools view -h -o "+syspathrs+"mapped"+str(k)+".sam "+syspathrs+"mapped"+str(k)+".bam") #get reversed mapped reads
77 os.system("cut -f 1 "+syspathrs+"unmapped"+str(k)+".sam > "+syspathrs+"unmapped"+str(k)+".txt")
78 os.system("cut -f 1 "+syspathrs+"mapped"+str(k)+".sam > "+syspathrs+"mapped"+str(k)+".txt")
79 os.system("python "+syspath+"remove_map.py "+syspathrs+"unmapped"+str(k)+".txt "+syspathrs+"mapped"+str(k)+".txt "+syspathrs+"runmapped"+str(k)+".txt")
80 os.system("rm "+syspathrs+"mapped"+str(k)+".sam")
81 os.system("rm "+syspathrs+"mapped"+str(k)+".txt")
82 os.system("rm "+syspathrs+"unmapped"+str(k)+".txt")
83 else:
84 os.system("cut -f 1 "+syspathrs+"unmapped"+str(k)+".sam > "+syspathrs+"runmapped"+str(k)+".txt")
85
86 os.system("rm "+syspathrs+"unmapped"+str(k)+".bam")
87 os.system("rm "+syspathrs+"umapped"+str(k)+".bam")
88 os.system("rm "+syspathrs+"rmapped"+str(k)+".bam")
89 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
90 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
91 os.system("rm "+syspathrs+"seq"+str(k)+".fa") #Remove sequences being mapped
92 os.system("rm "+syspathrs+"map"+str(k)+".sam") #Remove mapping file
93 os.system("rm "+syspathrs+"unmap_seq"+str(k)+".fa") #Remove unmapped sequnce
94 os.system("rm "+syspathrs+"runmapped"+str(k)+".txt")
95 os.system("rm "+syspathrs+"unmapped"+str(k)+".sam")
96
97 os.system("wc -l "+syspathrs+"seq"+str(k+1)+".fa > "+syspathrs+"count"+str(k+1)+".txt")
98 c = read_sp_file(syspathrs+"count"+str(k+1)+".txt")
99 if c[0][0] == '0': #If no reads is in the sequence file, stop
100 os.system("rm "+syspathrs+"count"+str(k+1)+".txt")
101 os.system("rm "+syspathrs+"seq"+str(k+1)+".fa")
102 break
103 os.system("rm "+syspathrs+"count"+str(k+1)+".txt")
104 k = k+1
105 if type_input == "fasta":
106 os.system("bowtie "+s+"-f "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam")
107 if type_input == "fastq":
108 os.system("bowtie "+s+"-q "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam")
109
110
111 ss = ""
112 for i in range(0,k+1):
113 ss = ss+" "+syspathrs+"mapped"+str(i)+".bam"
114
115
116 os.system("samtools merge -f "+output_file+" "+ss)
117 #print("samtools merge mapped_all.bam"+ss)
118 os.system("rm "+syspathrs+"mapped*.bam")
119 os.system("rm "+syspathrs+"ref*")
120 #os.system("rm -r "+syspathrs)
121
122