annotate breakdancer.py @ 10:00952c82ebc5 draft

Uploaded
author jeremie
date Fri, 27 Jun 2014 05:32:01 -0400
parents
children 0fb455068d3c
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
10
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
1 #!/usr/bin/python
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
2
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
3 import argparse, optparse, os, shutil, subprocess, sys, tempfile, shlex, time
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
4
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
5
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
6 parser = argparse.ArgumentParser(description='')
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
7
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
8 parser.add_argument ( '-i1', dest='input_bam', help='the bam input file' )
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
9 parser.add_argument ( '-o1', dest='output_raw', help='the output file' )
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
10 parser.add_argument ( '-o2', dest='output_vcf', help='the output file' )
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
11
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
12
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
13 tmp_dir = tempfile.mkdtemp()
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
14 errorFile = tmp_dir+"/errorLog"
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
15
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
16 bam2cfg_path = "bam2cfg.pl"
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
17 # breakdancer_path = "/usr/bin/breakdancer-max"
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
18 breakdancer_path = "breakdancer-max"
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
19 breakdancer2vcf_path = "breakdancer2vcf.py"
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
20
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
21
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
22 def bam2cfg(args):
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
23 config = tmp_dir+"/breakdancer_config"
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
24 cmd = 'perl %s %s' % (bam2cfg_path, args.input_bam)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
25 execute(cmd, output=config)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
26 print ("\ncmd = %s \n" ) %(cmd)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
27 return config
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
28
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
29
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
30 def breakdancer(args, config):
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
31 cmd = '%s %s' % (breakdancer_path, config)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
32 execute(cmd, output=args.output_raw)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
33
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
34
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
35 def breakdancer2vcf(args):
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
36 cmd = "python %s -i %s -o %s" % ( breakdancer2vcf_path, args.output_raw, args.output_vcf )
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
37 execute(cmd)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
38
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
39
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
40 def execute( cmd, output="" ):
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
41 try:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
42 err = open(tmp_dir+"/errorLog", 'a')
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
43 if output != "":
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
44 out = open(output, 'w')
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
45 else:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
46 out = subprocess.PIPE
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
47 process = subprocess.Popen( args=shlex.split(cmd), stdout=out, stderr=err )
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
48 process.wait()
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
49 err.close()
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
50 if out != subprocess.PIPE:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
51 out.close()
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
52 except Exception, e:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
53 sys.stderr.write("problem doing : %s\n" %(cmd))
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
54 sys.stderr.write( '%s\n\n' % str(e) )
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
55
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
56
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
57 def check( output ):
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
58 if (os.path.getsize(output)>0):
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
59 return True
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
60 else:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
61 sys.stderr.write('The output file is empty : %s\n' % (output))
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
62
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
63
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
64 def getLine(file):
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
65 try:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
66 f = open(file, 'r')
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
67 lignes = f.readlines()
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
68 n=0
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
69 for ligne in lignes:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
70 if ligne.strip()[0]!="#":
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
71 n+=1
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
72 # n = len(lignes)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
73 f.close()
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
74 except Exception, e:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
75 sys.stderr.write( '%s\n' % str(e) )
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
76 return n
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
77
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
78
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
79 def compare( output1, output2 ):
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
80 # compare le nombre de ligne entre 2 fichiers
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
81 # pour verifier qu'aucune ligne n'a ete saute
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
82 num_raw = getLine(output1)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
83 num_vcf = getLine(output2)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
84 if (num_raw==num_vcf):
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
85 return True
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
86 else:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
87 sys.stderr.write('Not the same number of variant between the raw file and the vcf file : %d vs %d\n' % (num_raw, num_vcf))
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
88
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
89
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
90 def __main__():
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
91
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
92 time0 = time.time()
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
93 args = parser.parse_args()
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
94
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
95 try:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
96 config = bam2cfg(args)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
97 breakdancer(args, config)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
98 breakdancer2vcf(args)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
99
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
100 # quelques tests
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
101 # verifier que les fichiers de sorties ne sont pas vides
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
102 check(args.output_raw)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
103 check(args.output_vcf)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
104 # comparer le nombre de ligne entre les 2 fichiers
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
105 # pour etre sur que toute les variations ont ete prises en compte
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
106 compare(args.output_raw, args.output_vcf)
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
107
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
108 sys.stdout.write( '\nDone in %d seconds\n' %(int(time.time()-time0)))
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
109
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
110 if ( os.path.getsize(errorFile)>0 ):
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
111 sys.stdout.write( 'At least one non fatal error was send, check the file : %s\n' %(errorFile) )
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
112 try:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
113 err = open(errorFile, 'r')
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
114 errors = err.read()
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
115 err.close()
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
116 sys.stdout.write( "Errors :\n%s" %(errors))
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
117 except Exception, e:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
118 sys.stderr.write( '%s\n' % str(e) )
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
119 else:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
120 sys.stdout.write( 'BreakDancer successful' )
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
121
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
122 except Exception, e:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
123 sys.stdout.write( 'BreakDancer fail\n' )
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
124 sys.stderr.write( '%s\n' % str(e) )
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
125 sys.exit()
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
126
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
127 finally:
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
128 if os.path.exists( tmp_dir ):
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
129 shutil.rmtree( tmp_dir )
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
130
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
131 if __name__=="__main__":
00952c82ebc5 Uploaded
jeremie
parents:
diff changeset
132 __main__()