Mercurial > repos > drosofff > msp_sr_bowtie

comparison sRbowtie.py @ 0:64064dccdb11 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Imported from capsule None
author drosofff
date Mon, 03 Nov 2014 10:24:38 -0500
parents
children b50d7228b678
comparison
equal deleted inserted replaced
-1:000000000000 0:64064dccdb11
1 #!/usr/bin/env python
2 # small RNA oriented bowtie wrapper
3 # version 1.5 17-7-2014: arg parser implementation
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}>
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
6 # Christophe Antoniewski <drosofff@gmail.com>
7
8 import sys, os, subprocess, tempfile, shutil, argparse
9
10 def Parser():
11 the_parser = argparse.ArgumentParser(description="bowtie wrapper for small fasta reads")
12 the_parser.add_argument('--input', action="store", type=str, help="input fasta file")
13 the_parser.add_argument('--method', action="store", type=str, help="RNA, unique, multiple, k_option, n_option, a_option")
14 the_parser.add_argument('--v-mismatches', dest="v_mismatches", action="store", type=str, help="number of mismatches allowed for the alignments")
15 the_parser.add_argument('--output-format', dest="output_format", action="store", type=str, help="tabular, sam, bam")
16 the_parser.add_argument('--output', action="store", type=str, help="output file path")
17 the_parser.add_argument('--index-from', dest="index_from", action="store", type=str, help="indexed or history")
18 the_parser.add_argument('--index-source', dest="index_source", action="store", type=str, help="file path to the index source")
19 the_parser.add_argument('--aligned', action="store", type=str, help="aligned read file path, maybe None")
20 the_parser.add_argument('--unaligned', action="store", type=str, help="unaligned read file path, maybe None")
21 the_parser.add_argument('--num-threads', dest="num_threads", action="store", type=str, help="number of bowtie threads")
22 args = the_parser.parse_args()
23 return args
24
25 def stop_err( msg ):
26 sys.stderr.write( '%s\n' % msg )
27 sys.exit()
28
29 def bowtieCommandLiner (alignment_method="RNA", v_mis="1", out_type="tabular", aligned="None", unaligned="None", input="path", index="path", output="path", pslots="4"):
30 if alignment_method=="RNA":
31 x = "-v %s -M 1 --best --strata -p %s --norc --suppress 6,7,8" % (v_mis, pslots)
32 elif alignment_method=="unique":
33 x = "-v %s -m 1 -p %s --suppress 6,7,8" % (v_mis, pslots)
34 elif alignment_method=="multiple":
35 x = "-v %s -M 1 --best --strata -p %s --suppress 6,7,8" % (v_mis, pslots)
36 elif alignment_method=="k_option":
37 x = "-v %s -k 1 --best -p %s --suppress 6,7,8" % (v_mis, pslots)
38 elif alignment_method=="n_option":
39 x = "-n %s -M 1 --best -p %s --suppress 6,7,8" % (v_mis, pslots)
40 elif alignment_method=="a_option":
41 x = "-v %s -a --best -p %s --suppress 6,7,8" % (v_mis, pslots)
42 if aligned == "None" and unaligned == "None": fasta_command = ""
43 elif aligned != "None" and unaligned == "None": fasta_command= " --al %s" % aligned
44 elif aligned == "None" and unaligned != "None": fasta_command = " --un %s" % unaligned
45 else: fasta_command = " --al %s --un %s" % (aligned, unaligned)
46 x = x + fasta_command
47 if out_type == "tabular":
48 return "bowtie %s %s -f %s > %s" % (x, index, input, output)
49 elif out_type=="sam":
50 return "bowtie %s -S %s -f %s > %s" % (x, index, input, output)
51 elif out_type=="bam":
52 return "bowtie %s -S %s -f %s |samtools view -bS - > %s" % (x, index, input, output)
53
54 def bowtie_squash(fasta):
55 tmp_index_dir = tempfile.mkdtemp() # make temp directory for bowtie indexes
56 ref_file = tempfile.NamedTemporaryFile( dir=tmp_index_dir )
57 ref_file_name = ref_file.name
58 ref_file.close() # by default, delete the temporary file, but ref_file.name is now stored in ref_file_name
59 os.symlink( fasta, ref_file_name ) # symlink between the fasta source file and the deleted ref_file name
60 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)
61 try:
62 FNULL = open(os.devnull, 'w')
63 tmp = tempfile.NamedTemporaryFile( dir=tmp_index_dir ).name # a path string for a temp file in tmp_index_dir. Just a string
64 tmp_stderr = open( tmp, 'wb' ) # creates and open a file handler pointing to the temp file
65 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
66 returncode = proc.wait()
67 tmp_stderr.close()
68 FNULL.close()
69 sys.stdout.write(cmd1 + "\n")
70 except Exception, e:
71 # clean up temp dir
72 if os.path.exists( tmp_index_dir ):
73 shutil.rmtree( tmp_index_dir )
74 stop_err( 'Error indexing reference sequence\n' + str( e ) )
75 # no Cleaning if no Exception, tmp_index_dir has to be cleaned after bowtie_alignment()
76 index_full_path = os.path.join(tmp_index_dir, ref_file_name) # bowtie fashion path without extention
77 return tmp_index_dir, index_full_path
78
79 def bowtie_alignment(command_line, flyPreIndexed=''):
80 # make temp directory just for stderr
81 tmp_index_dir = tempfile.mkdtemp()
82 tmp = tempfile.NamedTemporaryFile( dir=tmp_index_dir ).name
83 tmp_stderr = open( tmp, 'wb' )
84 # conditional statement for sorted bam generation viewable in Trackster
85 if "samtools" in command_line:
86 target_file = command_line.split()[-1] # recover the final output file name
87 path_to_unsortedBam = os.path.join(tmp_index_dir, "unsorted.bam")
88 path_to_sortedBam = os.path.join(tmp_index_dir, "unsorted.bam.sorted")
89 first_command_line = " ".join(command_line.split()[:-3]) + " -o " + path_to_unsortedBam + " - "
90 # 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 -
91 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
92 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
93 returncode = p.wait()
94 sys.stdout.write("%s\n" % first_command_line + str(returncode))
95 p = subprocess.Popen(args=second_command_line, cwd=tmp_index_dir, shell=True, stderr=tmp_stderr.fileno())
96 returncode = p.wait()
97 sys.stdout.write("\n%s\n" % second_command_line + str(returncode))
98 if os.path.isfile(path_to_sortedBam + ".bam"):
99 shutil.copy2(path_to_sortedBam + ".bam", target_file)
100 else:
101 p = subprocess.Popen(args=command_line, shell=True, stderr=tmp_stderr.fileno())
102 returncode = p.wait()
103 sys.stdout.write(command_line + "\n")
104 tmp_stderr.close()
105 ## cleaning if the index was created in the fly
106 if os.path.exists( flyPreIndexed ):
107 shutil.rmtree( flyPreIndexed )
108 # cleaning tmp files and directories
109 if os.path.exists( tmp_index_dir ):
110 shutil.rmtree( tmp_index_dir )
111 return
112
113 def __main__():
114 args = Parser()
115 F = open (args.output, "w")
116 if args.index_from == "history":
117 tmp_dir, index_path = bowtie_squash(args.index_source)
118 else:
119 tmp_dir, index_path = "dummy/dymmy", args.index_source
120 command_line = bowtieCommandLiner(args.method, args.v_mismatches, args.output_format, args.aligned, args.unaligned, args.input, index_path, args.output, args.num_threads)
121 bowtie_alignment(command_line, flyPreIndexed=tmp_dir)
122 F.close()
123 if __name__=="__main__": __main__()