annotate Iterative_mapping/iterative_map.py @ 69:bd02fcf26f7a draft

Uploaded wrapper that correctly handles structure prediction without constraints.
author devteam
date Mon, 24 Nov 2014 18:21:30 -0500
parents d2817a631a7b
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
66
d2817a631a7b Uploaded
tyty
parents:
diff changeset
1 #!/usr/bin/env python
d2817a631a7b Uploaded
tyty
parents:
diff changeset
2 # -*- coding: utf-8 -*-
d2817a631a7b Uploaded
tyty
parents:
diff changeset
3
d2817a631a7b Uploaded
tyty
parents:
diff changeset
4 import sys
d2817a631a7b Uploaded
tyty
parents:
diff changeset
5 import os
d2817a631a7b Uploaded
tyty
parents:
diff changeset
6 from read_file import *
d2817a631a7b Uploaded
tyty
parents:
diff changeset
7 from read_s_file import *
d2817a631a7b Uploaded
tyty
parents:
diff changeset
8 import random
d2817a631a7b Uploaded
tyty
parents:
diff changeset
9 import string
d2817a631a7b Uploaded
tyty
parents:
diff changeset
10
d2817a631a7b Uploaded
tyty
parents:
diff changeset
11 type_input = sys.argv[1]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
12 seq_file = sys.argv[2]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
13 ref_file = sys.argv[3]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
14 shift = sys.argv[4]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
15 length = sys.argv[5]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
16 t_end = sys.argv[6]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
17 map_type = sys.argv[7]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
18 output_file = sys.argv[8]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
19
d2817a631a7b Uploaded
tyty
parents:
diff changeset
20
d2817a631a7b Uploaded
tyty
parents:
diff changeset
21 if map_type!="default":
d2817a631a7b Uploaded
tyty
parents:
diff changeset
22 s = ""
d2817a631a7b Uploaded
tyty
parents:
diff changeset
23 sm = ""
d2817a631a7b Uploaded
tyty
parents:
diff changeset
24 s = s+"-v "+sys.argv[9]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
25 sm = sm+"-v "+sys.argv[9]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
26 sm = sm+" -5 "+sys.argv[10]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
27 sm = sm+" -3 "+sys.argv[11]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
28 s = s+" -k "+sys.argv[12]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
29 sm = sm+" -k "+sys.argv[12]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
30 if sys.argv[13]:
d2817a631a7b Uploaded
tyty
parents:
diff changeset
31 s = s+" -a"
d2817a631a7b Uploaded
tyty
parents:
diff changeset
32 sm = sm+" -a"
d2817a631a7b Uploaded
tyty
parents:
diff changeset
33 if int(sys.argv[14])>=1:
d2817a631a7b Uploaded
tyty
parents:
diff changeset
34 s = s+" -m "+sys.argv[14]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
35 sm = sm+" -m "+sys.argv[14]
d2817a631a7b Uploaded
tyty
parents:
diff changeset
36 if sys.argv[15]:
d2817a631a7b Uploaded
tyty
parents:
diff changeset
37 s = s+" --best --strata "
d2817a631a7b Uploaded
tyty
parents:
diff changeset
38 sm = sm+" --best --strata "
d2817a631a7b Uploaded
tyty
parents:
diff changeset
39
d2817a631a7b Uploaded
tyty
parents:
diff changeset
40 else:
d2817a631a7b Uploaded
tyty
parents:
diff changeset
41 s = "-v 3 -a --best --strata "
d2817a631a7b Uploaded
tyty
parents:
diff changeset
42 sm = "-v 3 -a --best --strata "
d2817a631a7b Uploaded
tyty
parents:
diff changeset
43
d2817a631a7b Uploaded
tyty
parents:
diff changeset
44 ospath = os.path.realpath(sys.argv[0])
d2817a631a7b Uploaded
tyty
parents:
diff changeset
45 ost = ospath.split('/')
d2817a631a7b Uploaded
tyty
parents:
diff changeset
46 syspath = ""
d2817a631a7b Uploaded
tyty
parents:
diff changeset
47 for i in range(len(ost)-1):
d2817a631a7b Uploaded
tyty
parents:
diff changeset
48 syspath = syspath+ost[i].strip()
d2817a631a7b Uploaded
tyty
parents:
diff changeset
49 syspath = syspath+'/'
d2817a631a7b Uploaded
tyty
parents:
diff changeset
50
d2817a631a7b Uploaded
tyty
parents:
diff changeset
51 syspathrs = os.getcwd()
d2817a631a7b Uploaded
tyty
parents:
diff changeset
52
d2817a631a7b Uploaded
tyty
parents:
diff changeset
53 os.system("bowtie-build -f "+ref_file+" "+syspathrs+"ref > "+syspathrs+"log.txt")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
54
d2817a631a7b Uploaded
tyty
parents:
diff changeset
55 os.system("cp "+seq_file+" "+syspathrs+"seq0.fa")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
56
d2817a631a7b Uploaded
tyty
parents:
diff changeset
57 if type_input == "fasta":
d2817a631a7b Uploaded
tyty
parents:
diff changeset
58 tp = 'fasta'
d2817a631a7b Uploaded
tyty
parents:
diff changeset
59 if type_input == "fastq":
d2817a631a7b Uploaded
tyty
parents:
diff changeset
60 tp = 'fastq'
d2817a631a7b Uploaded
tyty
parents:
diff changeset
61
d2817a631a7b Uploaded
tyty
parents:
diff changeset
62 k = 0
d2817a631a7b Uploaded
tyty
parents:
diff changeset
63
d2817a631a7b Uploaded
tyty
parents:
diff changeset
64 if type_input == "fasta":
d2817a631a7b Uploaded
tyty
parents:
diff changeset
65 os.system("bowtie "+sm+"-f "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
66 if type_input == "fastq":
d2817a631a7b Uploaded
tyty
parents:
diff changeset
67 os.system("bowtie "+sm+"-q "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
68
d2817a631a7b Uploaded
tyty
parents:
diff changeset
69 while(True):
d2817a631a7b Uploaded
tyty
parents:
diff changeset
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
d2817a631a7b Uploaded
tyty
parents:
diff changeset
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
d2817a631a7b Uploaded
tyty
parents:
diff changeset
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
d2817a631a7b Uploaded
tyty
parents:
diff changeset
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
d2817a631a7b Uploaded
tyty
parents:
diff changeset
74 os.system("samtools view -h -o "+syspathrs+"unmapped"+str(k)+".sam "+syspathrs+"unmapped"+str(k)+".bam") #get reversed mapped reads
d2817a631a7b Uploaded
tyty
parents:
diff changeset
75 if k>0:
d2817a631a7b Uploaded
tyty
parents:
diff changeset
76 os.system("samtools view -h -o "+syspathrs+"mapped"+str(k)+".sam "+syspathrs+"mapped"+str(k)+".bam") #get reversed mapped reads
d2817a631a7b Uploaded
tyty
parents:
diff changeset
77 os.system("cut -f 1 "+syspathrs+"unmapped"+str(k)+".sam > "+syspathrs+"unmapped"+str(k)+".txt")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
78 os.system("cut -f 1 "+syspathrs+"mapped"+str(k)+".sam > "+syspathrs+"mapped"+str(k)+".txt")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
79 os.system("python "+syspath+"remove_map.py "+syspathrs+"unmapped"+str(k)+".txt "+syspathrs+"mapped"+str(k)+".txt "+syspathrs+"runmapped"+str(k)+".txt")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
80 os.system("rm "+syspathrs+"mapped"+str(k)+".sam")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
81 os.system("rm "+syspathrs+"mapped"+str(k)+".txt")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
82 os.system("rm "+syspathrs+"unmapped"+str(k)+".txt")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
83 else:
d2817a631a7b Uploaded
tyty
parents:
diff changeset
84 os.system("cut -f 1 "+syspathrs+"unmapped"+str(k)+".sam > "+syspathrs+"runmapped"+str(k)+".txt")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
85
d2817a631a7b Uploaded
tyty
parents:
diff changeset
86 os.system("rm "+syspathrs+"unmapped"+str(k)+".bam")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
87 os.system("rm "+syspathrs+"umapped"+str(k)+".bam")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
88 os.system("rm "+syspathrs+"rmapped"+str(k)+".bam")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
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
d2817a631a7b Uploaded
tyty
parents:
diff changeset
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
d2817a631a7b Uploaded
tyty
parents:
diff changeset
91 os.system("rm "+syspathrs+"seq"+str(k)+".fa") #Remove sequences being mapped
d2817a631a7b Uploaded
tyty
parents:
diff changeset
92 os.system("rm "+syspathrs+"map"+str(k)+".sam") #Remove mapping file
d2817a631a7b Uploaded
tyty
parents:
diff changeset
93 os.system("rm "+syspathrs+"unmap_seq"+str(k)+".fa") #Remove unmapped sequnce
d2817a631a7b Uploaded
tyty
parents:
diff changeset
94 os.system("rm "+syspathrs+"runmapped"+str(k)+".txt")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
95 os.system("rm "+syspathrs+"unmapped"+str(k)+".sam")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
96
d2817a631a7b Uploaded
tyty
parents:
diff changeset
97 os.system("wc -l "+syspathrs+"seq"+str(k+1)+".fa > "+syspathrs+"count"+str(k+1)+".txt")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
98 c = read_sp_file(syspathrs+"count"+str(k+1)+".txt")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
99 if c[0][0] == '0': #If no reads is in the sequence file, stop
d2817a631a7b Uploaded
tyty
parents:
diff changeset
100 os.system("rm "+syspathrs+"count"+str(k+1)+".txt")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
101 os.system("rm "+syspathrs+"seq"+str(k+1)+".fa")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
102 break
d2817a631a7b Uploaded
tyty
parents:
diff changeset
103 os.system("rm "+syspathrs+"count"+str(k+1)+".txt")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
104 k = k+1
d2817a631a7b Uploaded
tyty
parents:
diff changeset
105 if type_input == "fasta":
d2817a631a7b Uploaded
tyty
parents:
diff changeset
106 os.system("bowtie "+s+"-f "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
107 if type_input == "fastq":
d2817a631a7b Uploaded
tyty
parents:
diff changeset
108 os.system("bowtie "+s+"-q "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
109
d2817a631a7b Uploaded
tyty
parents:
diff changeset
110
d2817a631a7b Uploaded
tyty
parents:
diff changeset
111 ss = ""
d2817a631a7b Uploaded
tyty
parents:
diff changeset
112 for i in range(0,k+1):
d2817a631a7b Uploaded
tyty
parents:
diff changeset
113 ss = ss+" "+syspathrs+"mapped"+str(i)+".bam"
d2817a631a7b Uploaded
tyty
parents:
diff changeset
114
d2817a631a7b Uploaded
tyty
parents:
diff changeset
115
d2817a631a7b Uploaded
tyty
parents:
diff changeset
116 os.system("samtools merge -f "+output_file+" "+ss)
d2817a631a7b Uploaded
tyty
parents:
diff changeset
117 #print("samtools merge mapped_all.bam"+ss)
d2817a631a7b Uploaded
tyty
parents:
diff changeset
118 os.system("rm "+syspathrs+"mapped*.bam")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
119 os.system("rm "+syspathrs+"ref*")
d2817a631a7b Uploaded
tyty
parents:
diff changeset
120 #os.system("rm -r "+syspathrs)
d2817a631a7b Uploaded
tyty
parents:
diff changeset
121
d2817a631a7b Uploaded
tyty
parents:
diff changeset
122