annotate DiffExpAnal/bam_to_sam_parallel.py @ 10:6e573fd3c41b draft

Uploaded
author yufei-luo
date Mon, 13 May 2013 10:06:30 -0400
parents 63799b789162
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
1 #!/usr/bin/env python
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
2 """
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
3 Yufei LUO
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
4 """
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
5 import optparse, os, sys, subprocess, tempfile, shutil, tarfile, random
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
6 #from galaxy import eggs
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
7 #import pkg_resources; pkg_resources.require( "bx-python" )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
8 #from bx.cookbook import doc_optparse
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
9 #from galaxy import util
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
10
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
11 def stop_err( msg ):
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
12 sys.stderr.write( '%s\n' % msg )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
13 sys.exit()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
14
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
15 def toTar(tarFileName, samOutputNames):
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
16 dir = os.path.dirname(tarFileName)
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
17 tfile = tarfile.open(tarFileName + ".tmp.tar", "w")
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
18 currentPath = os.getcwd()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
19 os.chdir(dir)
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
20 for file in samOutputNames:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
21 relativeFileName = os.path.basename(file)
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
22 tfile.add(relativeFileName)
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
23 os.system("mv %s %s" % (tarFileName + ".tmp.tar", tarFileName))
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
24 tfile.close()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
25 os.chdir(currentPath)
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
26
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
27
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
28 def __main__():
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
29 #Parse Command Line
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
30 parser = optparse.OptionParser()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
31 parser.add_option('-t', '--tar', dest='outputTar', default=None, help='output all SAM results in a tar file.' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
32 parser.add_option( '', '--input1', dest='input1', help='The input list of BAM datasets on txt format.' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
33 #parser.add_option( '', '--input1', dest='input1', help='The input BAM dataset' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
34 parser.add_option( '', '--output1', dest='output1', help='The output list of SAM datasets on txt format.' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
35 #parser.add_option( '', '--output1', dest='output1', help='The output SAM dataset' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
36 parser.add_option( '', '--header', dest='header', action='store_true', default=False, help='Write SAM Header' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
37 ( options, args ) = parser.parse_args()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
38
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
39
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
40 #Parse the input txt file and read a list of BAM files.
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
41 file = open(options.input1, "r")
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
42 lines = file.readlines()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
43 inputFileNames = []
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
44 samOutputNames = []
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
45 outputName = options.output1
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
46 resDirName = os.path.dirname(outputName) + '/'
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
47 #Write output txt file and define all output sam file names.
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
48 out = open(outputName, "w")
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
49 for line in lines:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
50 tab = line.split()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
51 inputFileNames.append(tab[1])
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
52 samOutName = resDirName + tab[0] + '_samOutput_%s.sam' % random.randrange(0, 10000)
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
53 samOutputNames.append(samOutName)
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
54 out.write(tab[0] + '\t' + samOutName + '\n')
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
55 file.close()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
56 out.close()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
57
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
58 # output version # of tool
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
59 try:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
60 tmp_files = []
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
61 tmp = tempfile.NamedTemporaryFile().name
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
62 tmp_files.append(tmp)
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
63 tmp_stdout = open( tmp, 'wb' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
64 proc = subprocess.Popen( args='samtools 2>&1', shell=True, stdout=tmp_stdout )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
65 tmp_stdout.close()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
66 returncode = proc.wait()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
67 stdout = None
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
68 for line in open( tmp_stdout.name, 'rb' ):
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
69 if line.lower().find( 'version' ) >= 0:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
70 stdout = line.strip()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
71 break
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
72 if stdout:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
73 sys.stdout.write( 'Samtools %s\n' % stdout )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
74 else:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
75 raise Exception
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
76 except:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
77 sys.stdout.write( 'Could not determine Samtools version\n' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
78
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
79
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
80
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
81 tmp_dirs = []
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
82 for i in range(len(inputFileNames)):
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
83 try:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
84 # exit if input file empty
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
85 if os.path.getsize( inputFileNames[i] ) == 0:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
86 raise Exception, 'Initial input txt file is empty.'
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
87 # Sort alignments by leftmost coordinates. File <out.prefix>.bam will be created. This command
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
88 # may also create temporary files <out.prefix>.%d.bam when the whole alignment cannot be fitted
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
89 # into memory ( controlled by option -m ).
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
90 tmp_dir = tempfile.mkdtemp()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
91 tmp_dirs.append(tmp_dir)
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
92 tmp_sorted_aligns_file = tempfile.NamedTemporaryFile( dir=tmp_dir )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
93 tmp_sorted_aligns_file_base = tmp_sorted_aligns_file.name
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
94 tmp_sorted_aligns_file_name = '%s.bam' % tmp_sorted_aligns_file.name
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
95 tmp_files.append(tmp_sorted_aligns_file_name)
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
96 tmp_sorted_aligns_file.close()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
97
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
98 command = 'samtools sort %s %s' % ( inputFileNames[i], tmp_sorted_aligns_file_base )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
99 tmp = tempfile.NamedTemporaryFile( dir=tmp_dir ).name
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
100 tmp_stderr = open( tmp, 'wb' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
101 proc = subprocess.Popen( args=command, shell=True, cwd=tmp_dir, stderr=tmp_stderr.fileno() )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
102 returncode = proc.wait()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
103 tmp_stderr.close()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
104 # get stderr, allowing for case where it's very large
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
105 tmp_stderr = open( tmp, 'rb' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
106 stderr = ''
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
107 buffsize = 1048576
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
108 try:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
109 while True:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
110 stderr += tmp_stderr.read( buffsize )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
111 if not stderr or len( stderr ) % buffsize != 0:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
112 break
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
113 except OverflowError:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
114 pass
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
115 tmp_stderr.close()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
116 if returncode != 0:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
117 raise Exception, stderr
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
118 # exit if sorted BAM file empty
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
119 if os.path.getsize( tmp_sorted_aligns_file_name) == 0:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
120 raise Exception, 'Intermediate sorted BAM file empty'
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
121 except Exception, e:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
122 stop_err( 'Error sorting alignments from (%s), %s' % ( inputFileNames[i], str( e ) ) )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
123
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
124 try:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
125 # Extract all alignments from the input BAM file to SAM format ( since no region is specified, all the alignments will be extracted ).
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
126 if options.header:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
127 view_options = "-h"
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
128 else:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
129 view_options = ""
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
130 command = 'samtools view %s -o %s %s' % ( view_options, samOutputNames[i], tmp_sorted_aligns_file_name )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
131 tmp = tempfile.NamedTemporaryFile( dir=tmp_dir ).name
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
132 tmp_stderr = open( tmp, 'wb' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
133 proc = subprocess.Popen( args=command, shell=True, cwd=tmp_dir, stderr=tmp_stderr.fileno() )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
134 returncode = proc.wait()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
135 tmp_stderr.close()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
136 # get stderr, allowing for case where it's very large
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
137 tmp_stderr = open( tmp, 'rb' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
138 stderr = ''
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
139 buffsize = 1048576
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
140 try:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
141 while True:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
142 stderr += tmp_stderr.read( buffsize )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
143 if not stderr or len( stderr ) % buffsize != 0:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
144 break
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
145 except OverflowError:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
146 pass
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
147 tmp_stderr.close()
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
148 if returncode != 0:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
149 raise Exception, stderr
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
150 except Exception, e:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
151 stop_err( 'Error extracting alignments from (%s), %s' % ( inputFileNames[i], str( e ) ) )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
152 if os.path.getsize( samOutputNames[i] ) > 0:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
153 sys.stdout.write( 'BAM file converted to SAM' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
154 else:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
155 stop_err( 'The output file is empty, there may be an error with your input file.' )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
156
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
157 if options.outputTar != None:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
158 toTar(options.outputTar, samOutputNames)
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
159 #clean up temp files
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
160 for tmp_dir in tmp_dirs:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
161 if os.path.exists( tmp_dir ):
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
162 shutil.rmtree( tmp_dir )
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
163 #print tmp_files
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
164 #for tmp in tmp_files:
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
165 # os.remove(tmp)
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
166
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
167
63799b789162 Uploaded
yufei-luo
parents:
diff changeset
168 if __name__=="__main__": __main__()