annotate sRbowtie_wrapper.py @ 1:eed2a141eb0c draft

Uploaded
author drosofff
date Sun, 11 May 2014 18:18:25 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
1 #!/usr/bin/env python
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
2 # generic bowtie wrapper for bowtie, small RNA oriented
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
3 # version 1 17-1-2014
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
4 # Usage sRbowtie_wrapper.py <1 input_fasta_file> <2 alignment method> <3 -v mismatches> <4 out_type> <5 buildIndexIfHistory> <6 fasta/bowtie index> <7 bowtie output> <8 ali_fasta> <9 unali_fasta>
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
5
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
6 import sys, re, os, subprocess, shlex, tempfile, shutil
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
7
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
8 def stop_err( msg ):
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
9 sys.stderr.write( '%s\n' % msg )
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
10 sys.exit()
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
11
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
12 def bowtieCommandLiner (alignment_method, v_mis, out_type, aligned, unaligned, input, index, output):
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
13 if alignment_method=="RNA":
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
14 x = "-v %s -M 1 --best --strata -p 12 --norc --suppress 2,6,7,8" % v_mis
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
15 elif alignment_method=="unique":
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
16 x = "-v %s -m 1 -p 12 --suppress 6,7,8" % v_mis
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
17 elif alignment_method=="multiple":
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
18 x = "-v %s -M 1 --best --strata -p 12 --suppress 6,7,8" % v_mis
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
19 elif alignment_method=="k_option":
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
20 x = "-v %s -k 1 --best -p 12 --suppress 6,7,8" % v_mis
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
21 elif alignment_method=="n_option":
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
22 x = "-n %s -M 1 --best -p 12 --suppress 6,7,8" % v_mis
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
23 elif alignment_method=="a_option":
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
24 x = "-v %s -a --best -p 12 --suppress 6,7,8" % v_mis
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
25 if aligned == "None" and unaligned == "None": fasta_command = ""
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
26 elif aligned != "None" and unaligned == "None": fasta_command= " --al %s" % aligned
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
27 elif aligned == "None" and unaligned != "None": fasta_command = " --un %s" % unaligned
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
28 else: fasta_command = " --al %s --un %s" % (aligned, unaligned)
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
29 x = x + fasta_command
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
30 if out_type == "tabular":
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
31 return "bowtie %s %s -f %s > %s" % (x, index, input, output)
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
32 elif out_type=="sam":
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
33 return "bowtie %s -S %s -f %s > %s" % (x, index, input, output)
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
34 elif out_type=="bam":
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
35 return "bowtie %s -S %s -f %s |samtools view -bS - > %s" % (x, index, input, output)
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
36
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
37 def bowtie_squash(fasta):
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
38 # make temp directory for placement of indices and copy reference file there if necessary
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
39 tmp_index_dir = tempfile.mkdtemp()
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
40 ref_file = tempfile.NamedTemporaryFile( dir=tmp_index_dir )
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
41 ref_file_name = ref_file.name
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
42 ref_file.close() # by default, this action delete the temporary file !
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
43 os.symlink( fasta, ref_file_name ) # now there is a symlink between the fasta source file and the deleted ref_file name
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
44 cmd1 = 'bowtie-build -f %s %s' % (ref_file_name, ref_file_name ) # this will work with the bowtie command line but we have to change the working dir !
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
45 try:
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
46 FNULL = open(os.devnull, 'w')
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
47 tmp = tempfile.NamedTemporaryFile( dir=tmp_index_dir ).name # a full path temp file to tmp_index_dir is created. just a string
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
48 tmp_stderr = open( tmp, 'wb' ) # creates and open the file handler
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
49 proc = subprocess.Popen( args=cmd1, shell=True, cwd=tmp_index_dir, stderr=FNULL, stdout=FNULL ) # fileno() method return the file descriptor number of tmp_stderr / stderr=tmp_stderr.fileno()
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
50 # here I played a while a finish by redirecting everythin in dev/null. Clean later tmp_stderr calls
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
51 returncode = proc.wait()
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
52 tmp_stderr.close()
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
53 FNULL.close()
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
54 sys.stdout.write(cmd1 + "\n")
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
55 except Exception, e:
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
56 # clean up temp dir
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
57 if os.path.exists( tmp_index_dir ):
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
58 shutil.rmtree( tmp_index_dir )
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
59 stop_err( 'Error indexing reference sequence\n' + str( e ) )
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
60 # no Cleaning if no Exception, to be cleaned later after bowtie alignment
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
61 index_full_path = os.path.join(tmp_index_dir, ref_file_name) # bowtie fashion path (without extention) ...
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
62 return tmp_index_dir, index_full_path
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
63
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
64 def bowtie_alignment(command_line, flyPreIndexed=''):
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
65 # make temp directory just for stderr
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
66 tmp_index_dir = tempfile.mkdtemp()
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
67 tmp = tempfile.NamedTemporaryFile( dir=tmp_index_dir ).name
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
68 tmp_stderr = open( tmp, 'wb' )
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
69 # conditional statement for sorted bam generation viewable in Trackster
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
70 if "samtools" in command_line:
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
71 target_file = command_line.split()[-1] # recover the final output file name
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
72 path_to_unsortedBam = os.path.join(tmp_index_dir, "unsorted.bam")
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
73 path_to_sortedBam = os.path.join(tmp_index_dir, "unsorted.bam.sorted")
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
74 first_command_line = " ".join(command_line.split()[:-3]) + " -o " + path_to_unsortedBam + " - " # example: bowtie -v 0 -M 1 --best --strata -p 12 --suppress 6,7,8 -S /home/galaxy/galaxy-dist/bowtie/Dmel/dmel-all-chromosome-r5.49 -f /home/galaxy/galaxy-dist/database/files/003/dataset_3460.dat |samtools view -bS -o /tmp/tmp_PgMT0/unsorted.bam -
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
75 second_command_line = "samtools sort %s %s" % (path_to_unsortedBam, path_to_sortedBam) # Be carreful : this indeed will generate an unsorted.bam.sorted.bam file, NOT a unsorted.bam.sorted file
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
76 p = subprocess.Popen(args=first_command_line, cwd=tmp_index_dir, shell=True, stderr=tmp_stderr.fileno())
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
77 returncode = p.wait()
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
78 sys.stdout.write("%s\n" % first_command_line + str(returncode))
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
79 p = subprocess.Popen(args=second_command_line, cwd=tmp_index_dir, shell=True, stderr=tmp_stderr.fileno())
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
80 returncode = p.wait()
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
81 sys.stdout.write("\n%s\n" % second_command_line + str(returncode))
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
82 if os.path.isfile(path_to_sortedBam + ".bam"):
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
83 shutil.copy2(path_to_sortedBam + ".bam", target_file)
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
84 else:
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
85 p = subprocess.Popen(args=command_line, shell=True, stderr=tmp_stderr.fileno())
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
86 returncode = p.wait()
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
87 sys.stdout.write(command_line + "\n")
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
88 tmp_stderr.close()
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
89 ## cleaning if the index was created in the fly
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
90 if os.path.exists( flyPreIndexed ):
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
91 shutil.rmtree( flyPreIndexed )
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
92 # cleaning tmp files and directories
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
93 if os.path.exists( tmp_index_dir ):
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
94 shutil.rmtree( tmp_index_dir )
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
95 return
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
96
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
97 def __main__():
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
98 F = open (sys.argv[-3], "w")
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
99 if sys.argv[5] == "history":
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
100 tmp_dir, index_path = bowtie_squash(sys.argv[6])
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
101 else:
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
102 tmp_dir, index_path = "dummy/dymmy", sys.argv[6]
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
103 command_line = bowtieCommandLiner(sys.argv[2], sys.argv[3], sys.argv[4], sys.argv[-2], sys.argv[-1], sys.argv[1], index_path, sys.argv[7])
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
104 bowtie_alignment(command_line, flyPreIndexed=tmp_dir)
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
105 F.close()
eed2a141eb0c Uploaded
drosofff
parents:
diff changeset
106 if __name__=="__main__": __main__()