Mercurial > repos > drosofff > msp_sr_bowtie

comparison sRbowtie.py @ 0:e8bdae1a2bdc draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author drosofff
date Tue, 26 May 2015 18:36:09 -0400
parents
children 615d2550977f
comparison
equal deleted inserted replaced
-1:000000000000 0:e8bdae1a2bdc
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
9 import os
10 import subprocess
11 import tempfile
12 import shutil
13 import argparse
14
15
16 def Parser():
17 the_parser = argparse.ArgumentParser(
18 description="bowtie wrapper for small fasta reads")
19 the_parser.add_argument(
20 '--input', action="store", type=str, help="input file")
21 the_parser.add_argument(
22 '--input-format', dest="input_format", action="store", type=str, help="fasta or fastq")
23 the_parser.add_argument('--method', action="store", type=str,
24 help="RNA, unique, multiple, k_option, n_option, a_option")
25 the_parser.add_argument('--v-mismatches', dest="v_mismatches", action="store",
26 type=str, help="number of mismatches allowed for the alignments")
27 the_parser.add_argument(
28 '--output-format', dest="output_format", action="store", type=str, help="tabular, sam, bam")
29 the_parser.add_argument(
30 '--output', action="store", type=str, help="output file path")
31 the_parser.add_argument(
32 '--index-from', dest="index_from", action="store", type=str, help="indexed or history")
33 the_parser.add_argument('--index-source', dest="index_source",
34 action="store", type=str, help="file path to the index source")
35 the_parser.add_argument(
36 '--aligned', action="store", type=str, help="aligned read file path, maybe None")
37 the_parser.add_argument('--unaligned', action="store",
38 type=str, help="unaligned read file path, maybe None")
39 the_parser.add_argument('--num-threads', dest="num_threads",
40 action="store", type=str, help="number of bowtie threads")
41 args = the_parser.parse_args()
42 return args
43
44
45 def stop_err(msg):
46 sys.stderr.write('%s\n' % msg)
47 sys.exit()
48
49
50 def bowtieCommandLiner(alignment_method="RNA", v_mis="1", out_type="tabular",
51 aligned="None", unaligned="None", input_format="fasta", input="path",
52 index="path", output="path", pslots="4"):
53 if input_format == "fasta":
54 input_format = "-f"
55 elif (input_format == "fastq") or (input_format == "fastqsanger"):
56 input_format = "-q"
57 else:
58 raise Exception('input format must be one of fasta or fastq')
59 if alignment_method == "RNA":
60 x = "-v %s -M 1 --best --strata -p %s --norc --suppress 6,7,8" % (
61 v_mis, pslots)
62 elif alignment_method == "unique":
63 x = "-v %s -m 1 -p %s --suppress 6,7,8" % (v_mis, pslots)
64 elif alignment_method == "multiple":
65 x = "-v %s -M 1 --best --strata -p %s --suppress 6,7,8" % (
66 v_mis, pslots)
67 elif alignment_method == "k_option":
68 x = "-v %s -k 1 --best -p %s --suppress 6,7,8" % (v_mis, pslots)
69 elif alignment_method == "n_option":
70 x = "-n %s -M 1 --best -p %s --suppress 6,7,8" % (v_mis, pslots)
71 elif alignment_method == "a_option":
72 x = "-v %s -a --best -p %s --suppress 6,7,8" % (v_mis, pslots)
73 if aligned == "None" and unaligned == "None":
74 fasta_command = ""
75 elif aligned != "None" and unaligned == "None":
76 fasta_command = " --al %s" % aligned
77 elif aligned == "None" and unaligned != "None":
78 fasta_command = " --un %s" % unaligned
79 else:
80 fasta_command = " --al %s --un %s" % (aligned, unaligned)
81 x = x + fasta_command
82 if out_type == "tabular":
83 return "bowtie %s %s %s %s > %s" % (x, index, input_format, input, output)
84 elif out_type == "sam":
85 return "bowtie %s -S %s %s %s > %s" % (x, index, input_format, input, output)
86 elif out_type == "bam":
87 return "bowtie %s -S %s %s %s |samtools view -bS - > %s" % (
88 x, index, input_format, input, output)
89
90
91 def bowtie_squash(fasta):
92 # make temp directory for bowtie indexes
93 tmp_index_dir = tempfile.mkdtemp()
94 ref_file = tempfile.NamedTemporaryFile(dir=tmp_index_dir)
95 ref_file_name = ref_file.name
96 # by default, delete the temporary file, but ref_file.name is now stored
97 # in ref_file_name
98 ref_file.close()
99 # symlink between the fasta source file and the deleted ref_file name
100 os.symlink(fasta, ref_file_name)
101 # bowtie command line, which will work after changing dir
102 # (cwd=tmp_index_dir)
103 cmd1 = 'bowtie-build -f %s %s' % (ref_file_name, ref_file_name)
104 try:
105 FNULL = open(os.devnull, 'w')
106 # a path string for a temp file in tmp_index_dir. Just a string
107 tmp = tempfile.NamedTemporaryFile(dir=tmp_index_dir).name
108 # creates and open a file handler pointing to the temp file
109 tmp_stderr = open(tmp, 'wb')
110 # both stderr and stdout of bowtie-build are redirected in dev/null
111 proc = subprocess.Popen(
112 args=cmd1, shell=True, cwd=tmp_index_dir, stderr=FNULL, stdout=FNULL)
113 returncode = proc.wait()
114 tmp_stderr.close()
115 FNULL.close()
116 sys.stdout.write(cmd1 + "\n")
117 except Exception as e:
118 # clean up temp dir
119 if os.path.exists(tmp_index_dir):
120 shutil.rmtree(tmp_index_dir)
121 stop_err('Error indexing reference sequence\n' + str(e))
122 # no Cleaning if no Exception, tmp_index_dir has to be cleaned after
123 # bowtie_alignment()
124 # bowtie fashion path without extention
125 index_full_path = os.path.join(tmp_index_dir, ref_file_name)
126 return tmp_index_dir, index_full_path
127
128
129 def bowtie_alignment(command_line, flyPreIndexed=''):
130 # make temp directory just for stderr
131 tmp_index_dir = tempfile.mkdtemp()
132 tmp = tempfile.NamedTemporaryFile(dir=tmp_index_dir).name
133 tmp_stderr = open(tmp, 'wb')
134 # conditional statement for sorted bam generation viewable in Trackster
135 if "samtools" in command_line:
136 # recover the final output file name
137 target_file = command_line.split()[-1]
138 path_to_unsortedBam = os.path.join(tmp_index_dir, "unsorted.bam")
139 path_to_sortedBam = os.path.join(tmp_index_dir, "unsorted.bam.sorted")
140 first_command_line = " ".join(
141 command_line.split()[:-3]) + " -o " + path_to_unsortedBam + " - "
142 # example: bowtie -v 0 -M 1 --best --strata -p 12 --suppress 6,7,8 -S
143 # /home/galaxy/galaxy-dist/bowtie/Dmel/dmel-all-chromosome-r5.49 -f
144 # /home/galaxy/galaxy-dist/database/files/003/dataset_3460.dat
145 # |samtools view -bS -o /tmp/tmp_PgMT0/unsorted.bam -
146 # generates an "unsorted.bam.sorted.bam file", NOT an
147 # "unsorted.bam.sorted" file
148 second_command_line = "samtools sort %s %s" % (
149 path_to_unsortedBam, path_to_sortedBam)
150 # fileno() method return the file descriptor number of tmp_stderr
151 p = subprocess.Popen(
152 args=first_command_line, cwd=tmp_index_dir, shell=True, stderr=tmp_stderr.fileno())
153 returncode = p.wait()
154 sys.stdout.write("%s\n" % first_command_line + str(returncode))
155 p = subprocess.Popen(
156 args=second_command_line, cwd=tmp_index_dir, shell=True, stderr=tmp_stderr.fileno())
157 returncode = p.wait()
158 sys.stdout.write("\n%s\n" % second_command_line + str(returncode))
159 if os.path.isfile(path_to_sortedBam + ".bam"):
160 shutil.copy2(path_to_sortedBam + ".bam", target_file)
161 else:
162 p = subprocess.Popen(
163 args=command_line, shell=True, stderr=tmp_stderr.fileno())
164 returncode = p.wait()
165 sys.stdout.write(command_line + "\n")
166 tmp_stderr.close()
167 # cleaning if the index was created in the fly
168 if os.path.exists(flyPreIndexed):
169 shutil.rmtree(flyPreIndexed)
170 # cleaning tmp files and directories
171 if os.path.exists(tmp_index_dir):
172 shutil.rmtree(tmp_index_dir)
173 return
174
175
176 def __main__():
177 args = Parser()
178 F = open(args.output, "w")
179 if args.index_from == "history":
180 tmp_dir, index_path = bowtie_squash(args.index_source)
181 else:
182 tmp_dir, index_path = "dummy/dymmy", args.index_source
183 command_line = bowtieCommandLiner(args.method, args.v_mismatches, args.output_format,
184 args.aligned, args.unaligned, args.input_format, args.input,
185 index_path, args.output, args.num_threads)
186 bowtie_alignment(command_line, flyPreIndexed=tmp_dir)
187 F.close()
188 if __name__ == "__main__":
189 __main__()