annotate sRbowtie/sRbowtie.py @ 4:303baef87cb7 draft

Uploaded
author drosofff
date Sat, 21 Jun 2014 09:54:35 -0400
parents 324c3142ca0f
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
1 #!/usr/bin/env python
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
2 # small RNA oriented bowtie wrapper
4
303baef87cb7 Uploaded
drosofff
parents: 0
diff changeset
3 # version 1.01 29-5-2014
0
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
4 # Usage sRbowtie.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> <10 --num-threads \${GALAXY_SLOTS:-4}>
4
303baef87cb7 Uploaded
drosofff
parents: 0
diff changeset
5 # current rev: for bowtie __norc, move from --supress 2,6,7,8 to --supress 6,7,8. Future Parser must be updated to take into account this standardisation
0
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
6 # To Do:
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
7 # implement an arg parser
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
8 # Christophe Antoniewski <drosofff@gmail.com>
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
9
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
10 import sys, os, subprocess, tempfile, shutil
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
11
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
12 def stop_err( msg ):
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
13 sys.stderr.write( '%s\n' % msg )
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
14 sys.exit()
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
15
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
16 def bowtieCommandLiner (alignment_method, v_mis, out_type, aligned, unaligned, input, index, output, pslots="12"):
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
17 if alignment_method=="RNA":
4
303baef87cb7 Uploaded
drosofff
parents: 0
diff changeset
18 x = "-v %s -M 1 --best --strata -p %s --norc --suppress 6,7,8" % (v_mis, pslots)
0
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
19 elif alignment_method=="unique":
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
20 x = "-v %s -m 1 -p %s --suppress 6,7,8" % (v_mis, pslots)
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
21 elif alignment_method=="multiple":
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
22 x = "-v %s -M 1 --best --strata -p %s --suppress 6,7,8" % (v_mis, pslots)
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
23 elif alignment_method=="k_option":
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
24 x = "-v %s -k 1 --best -p %s --suppress 6,7,8" % (v_mis, pslots)
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
25 elif alignment_method=="n_option":
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
26 x = "-n %s -M 1 --best -p %s --suppress 6,7,8" % (v_mis, pslots)
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
27 elif alignment_method=="a_option":
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
28 x = "-v %s -a --best -p %s --suppress 6,7,8" % (v_mis, pslots)
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
29 if aligned == "None" and unaligned == "None": fasta_command = ""
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
30 elif aligned != "None" and unaligned == "None": fasta_command= " --al %s" % aligned
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
31 elif aligned == "None" and unaligned != "None": fasta_command = " --un %s" % unaligned
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
32 else: fasta_command = " --al %s --un %s" % (aligned, unaligned)
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
33 x = x + fasta_command
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
34 if out_type == "tabular":
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
35 return "bowtie %s %s -f %s > %s" % (x, index, input, output)
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
36 elif out_type=="sam":
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
37 return "bowtie %s -S %s -f %s > %s" % (x, index, input, output)
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
38 elif out_type=="bam":
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
39 return "bowtie %s -S %s -f %s |samtools view -bS - > %s" % (x, index, input, output)
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
40
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
41 def bowtie_squash(fasta):
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
42 tmp_index_dir = tempfile.mkdtemp() # make temp directory for bowtie indexes
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
43 ref_file = tempfile.NamedTemporaryFile( dir=tmp_index_dir )
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
44 ref_file_name = ref_file.name
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
45 ref_file.close() # by default, delete the temporary file, but ref_file.name is now stored in ref_file_name
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
46 os.symlink( fasta, ref_file_name ) # symlink between the fasta source file and the deleted ref_file name
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
47 cmd1 = 'bowtie-build -f %s %s' % (ref_file_name, ref_file_name ) # bowtie command line, which will work after changing dir (cwd=tmp_index_dir)
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
48 try:
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
49 FNULL = open(os.devnull, 'w')
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
50 tmp = tempfile.NamedTemporaryFile( dir=tmp_index_dir ).name # a path string for a temp file in tmp_index_dir. Just a string
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
51 tmp_stderr = open( tmp, 'wb' ) # creates and open a file handler pointing to the temp file
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
52 proc = subprocess.Popen( args=cmd1, shell=True, cwd=tmp_index_dir, stderr=FNULL, stdout=FNULL ) # both stderr and stdout of bowtie-build are redirected in dev/null
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
53 returncode = proc.wait()
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
54 tmp_stderr.close()
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
55 FNULL.close()
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
56 sys.stdout.write(cmd1 + "\n")
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
57 except Exception, e:
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
58 # clean up temp dir
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
59 if os.path.exists( tmp_index_dir ):
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
60 shutil.rmtree( tmp_index_dir )
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
61 stop_err( 'Error indexing reference sequence\n' + str( e ) )
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
62 # no Cleaning if no Exception, tmp_index_dir has to be cleaned after bowtie_alignment()
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
63 index_full_path = os.path.join(tmp_index_dir, ref_file_name) # bowtie fashion path without extention
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
64 return tmp_index_dir, index_full_path
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
65
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
66 def bowtie_alignment(command_line, flyPreIndexed=''):
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
67 # make temp directory just for stderr
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
68 tmp_index_dir = tempfile.mkdtemp()
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
69 tmp = tempfile.NamedTemporaryFile( dir=tmp_index_dir ).name
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
70 tmp_stderr = open( tmp, 'wb' )
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
71 # conditional statement for sorted bam generation viewable in Trackster
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
72 if "samtools" in command_line:
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
73 target_file = command_line.split()[-1] # recover the final output file name
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
74 path_to_unsortedBam = os.path.join(tmp_index_dir, "unsorted.bam")
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
75 path_to_sortedBam = os.path.join(tmp_index_dir, "unsorted.bam.sorted")
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
76 first_command_line = " ".join(command_line.split()[:-3]) + " -o " + path_to_unsortedBam + " - "
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
77 # 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 -
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
78 second_command_line = "samtools sort %s %s" % (path_to_unsortedBam, path_to_sortedBam) # generates an "unsorted.bam.sorted.bam file", NOT an "unsorted.bam.sorted" file
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
79 p = subprocess.Popen(args=first_command_line, cwd=tmp_index_dir, shell=True, stderr=tmp_stderr.fileno()) # fileno() method return the file descriptor number of tmp_stderr
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
80 returncode = p.wait()
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
81 sys.stdout.write("%s\n" % first_command_line + str(returncode))
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
82 p = subprocess.Popen(args=second_command_line, cwd=tmp_index_dir, shell=True, stderr=tmp_stderr.fileno())
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
83 returncode = p.wait()
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
84 sys.stdout.write("\n%s\n" % second_command_line + str(returncode))
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
85 if os.path.isfile(path_to_sortedBam + ".bam"):
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
86 shutil.copy2(path_to_sortedBam + ".bam", target_file)
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
87 else:
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
88 p = subprocess.Popen(args=command_line, shell=True, stderr=tmp_stderr.fileno())
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
89 returncode = p.wait()
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
90 sys.stdout.write(command_line + "\n")
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
91 tmp_stderr.close()
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
92 ## cleaning if the index was created in the fly
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
93 if os.path.exists( flyPreIndexed ):
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
94 shutil.rmtree( flyPreIndexed )
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
95 # cleaning tmp files and directories
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
96 if os.path.exists( tmp_index_dir ):
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
97 shutil.rmtree( tmp_index_dir )
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
98 return
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
99
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
100 def __main__():
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
101 F = open (sys.argv[7], "w")
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
102 if sys.argv[5] == "history":
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
103 tmp_dir, index_path = bowtie_squash(sys.argv[6])
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
104 else:
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
105 tmp_dir, index_path = "dummy/dymmy", sys.argv[6]
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
106 command_line = bowtieCommandLiner(sys.argv[2], sys.argv[3], sys.argv[4], sys.argv[8], sys.argv[9], sys.argv[1], index_path, sys.argv[7], sys.argv[10])
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
107 bowtie_alignment(command_line, flyPreIndexed=tmp_dir)
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
108 F.close()
324c3142ca0f Uploaded
drosofff
parents:
diff changeset
109 if __name__=="__main__": __main__()