Mercurial > repos > tyty > structurefold
annotate Iterative_mapping/iterative_map.py @ 87:5a9183ec344f draft
Uploaded
author | tyty |
---|---|
date | Fri, 19 Dec 2014 13:23:03 -0500 (2014-12-19) |
parents | b6bdd41440f3 |
children | 87ec0ecdc2af |
rev | line source |
---|---|
71 | 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 syspathrs = syspathrs+'/' | |
53 | |
54 os.system("bowtie-build -f "+ref_file+" "+syspathrs+"ref > "+syspathrs+"log.txt") | |
55 | |
56 os.system("cp "+seq_file+" "+syspathrs+"seq0.fa") | |
57 | |
58 if type_input == "fasta": | |
59 tp = 'fasta' | |
60 if type_input == "fastq": | |
61 tp = 'fastq' | |
62 | |
63 k = 0 | |
64 | |
65 if type_input == "fasta": | |
66 os.system("bowtie "+sm+"-f "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam") | |
67 if type_input == "fastq": | |
68 os.system("bowtie "+sm+"-q "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam") | |
69 | |
70 while(True): | |
71 os.system("samtools view -Sb -F 0xfff "+syspathrs+"map"+str(k)+".sam > "+syspathrs+"mapped"+str(k)+".bam 2>"+syspathrs+"log.txt") #get mapped reads | |
72 os.system("samtools view -Sb -f 0x4 "+syspathrs+"map"+str(k)+".sam > "+syspathrs+"umapped"+str(k)+".bam 2>"+syspathrs+"log.txt") #get unmapped reads | |
73 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 | |
74 os.system("samtools merge -f "+syspathrs+"unmapped"+str(k)+".bam "+syspathrs+"umapped"+str(k)+".bam "+syspathrs+"rmapped"+str(k)+".bam") #get reversed mapped reads | |
75 os.system("samtools view -h -o "+syspathrs+"unmapped"+str(k)+".sam "+syspathrs+"unmapped"+str(k)+".bam") #get reversed mapped reads | |
76 if k>0: | |
77 os.system("samtools view -h -o "+syspathrs+"mapped"+str(k)+".sam "+syspathrs+"mapped"+str(k)+".bam") #get reversed mapped reads | |
78 os.system("cut -f 1 "+syspathrs+"unmapped"+str(k)+".sam > "+syspathrs+"unmapped"+str(k)+".txt") | |
79 os.system("cut -f 1 "+syspathrs+"mapped"+str(k)+".sam > "+syspathrs+"mapped"+str(k)+".txt") | |
80 os.system("python "+syspath+"remove_map.py "+syspathrs+"unmapped"+str(k)+".txt "+syspathrs+"mapped"+str(k)+".txt "+syspathrs+"runmapped"+str(k)+".txt") | |
81 os.system("rm "+syspathrs+"mapped"+str(k)+".sam") | |
82 os.system("rm "+syspathrs+"mapped"+str(k)+".txt") | |
83 os.system("rm "+syspathrs+"unmapped"+str(k)+".txt") | |
84 else: | |
85 os.system("cut -f 1 "+syspathrs+"unmapped"+str(k)+".sam > "+syspathrs+"runmapped"+str(k)+".txt") | |
86 | |
87 os.system("rm "+syspathrs+"unmapped"+str(k)+".bam") | |
88 os.system("rm "+syspathrs+"umapped"+str(k)+".bam") | |
89 os.system("rm "+syspathrs+"rmapped"+str(k)+".bam") | |
90 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 | |
91 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 | |
92 os.system("rm "+syspathrs+"seq"+str(k)+".fa") #Remove sequences being mapped | |
93 os.system("rm "+syspathrs+"map"+str(k)+".sam") #Remove mapping file | |
94 os.system("rm "+syspathrs+"unmap_seq"+str(k)+".fa") #Remove unmapped sequnce | |
95 os.system("rm "+syspathrs+"runmapped"+str(k)+".txt") | |
96 os.system("rm "+syspathrs+"unmapped"+str(k)+".sam") | |
97 | |
98 os.system("wc -l "+syspathrs+"seq"+str(k+1)+".fa > "+syspathrs+"count"+str(k+1)+".txt") | |
99 c = read_sp_file(syspathrs+"count"+str(k+1)+".txt") | |
100 if c[0][0] == '0': #If no reads is in the sequence file, stop | |
101 os.system("rm "+syspathrs+"count"+str(k+1)+".txt") | |
102 os.system("rm "+syspathrs+"seq"+str(k+1)+".fa") | |
103 break | |
104 os.system("rm "+syspathrs+"count"+str(k+1)+".txt") | |
105 k = k+1 | |
106 if type_input == "fasta": | |
107 os.system("bowtie "+s+"-f "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam") | |
108 if type_input == "fastq": | |
109 os.system("bowtie "+s+"-q "+syspathrs+"ref"+" "+syspathrs+"seq"+str(k)+".fa --quiet -S > "+syspathrs+"map"+str(k)+".sam") | |
110 | |
111 | |
112 ss = "" | |
113 for i in range(0,k+1): | |
114 ss = ss+" "+syspathrs+"mapped"+str(i)+".bam" | |
115 | |
116 | |
117 os.system("samtools merge -f "+syspathrs+"combine.bam"+" "+ss) | |
118 os.system("samtools sort "+syspathrs+"combine.bam sorted") | |
119 os.system("samtools view -b -h sorted.bam > " + output_file) | |
120 #print("samtools merge mapped_all.bam"+ss) | |
121 os.system("rm "+syspathrs+"mapped*.bam") | |
122 os.system("rm "+syspathrs+"combine.bam") | |
123 os.system("rm "+syspathrs+"sorted.bam") | |
124 os.system("rm "+syspathrs+"ref*") | |
125 #os.system("rm -r "+syspathrs) | |
126 | |
127 |