annotate sRbowtie.py @ 0:e8bdae1a2bdc draft

Uploaded
author drosofff
date Tue, 26 May 2015 18:36:09 -0400
parents
children 615d2550977f
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
1 #!/usr/bin/env python
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
2 # small RNA oriented bowtie wrapper
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
3 # version 1.5 17-7-2014: arg parser implementation
e8bdae1a2bdc 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}>
e8bdae1a2bdc Uploaded
drosofff
parents:
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
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
6 # Christophe Antoniewski <drosofff@gmail.com>
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
7
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
8 import sys
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
9 import os
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
10 import subprocess
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
11 import tempfile
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
12 import shutil
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
13 import argparse
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
14
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
15
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
16 def Parser():
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
17 the_parser = argparse.ArgumentParser(
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
18 description="bowtie wrapper for small fasta reads")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
19 the_parser.add_argument(
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
20 '--input', action="store", type=str, help="input file")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
21 the_parser.add_argument(
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
22 '--input-format', dest="input_format", action="store", type=str, help="fasta or fastq")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
23 the_parser.add_argument('--method', action="store", type=str,
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
24 help="RNA, unique, multiple, k_option, n_option, a_option")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
25 the_parser.add_argument('--v-mismatches', dest="v_mismatches", action="store",
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
26 type=str, help="number of mismatches allowed for the alignments")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
27 the_parser.add_argument(
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
28 '--output-format', dest="output_format", action="store", type=str, help="tabular, sam, bam")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
29 the_parser.add_argument(
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
30 '--output', action="store", type=str, help="output file path")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
31 the_parser.add_argument(
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
32 '--index-from', dest="index_from", action="store", type=str, help="indexed or history")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
33 the_parser.add_argument('--index-source', dest="index_source",
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
34 action="store", type=str, help="file path to the index source")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
35 the_parser.add_argument(
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
36 '--aligned', action="store", type=str, help="aligned read file path, maybe None")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
37 the_parser.add_argument('--unaligned', action="store",
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
38 type=str, help="unaligned read file path, maybe None")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
39 the_parser.add_argument('--num-threads', dest="num_threads",
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
40 action="store", type=str, help="number of bowtie threads")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
41 args = the_parser.parse_args()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
42 return args
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
43
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
44
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
45 def stop_err(msg):
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
46 sys.stderr.write('%s\n' % msg)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
47 sys.exit()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
48
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
49
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
50 def bowtieCommandLiner(alignment_method="RNA", v_mis="1", out_type="tabular",
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
51 aligned="None", unaligned="None", input_format="fasta", input="path",
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
52 index="path", output="path", pslots="4"):
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
53 if input_format == "fasta":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
54 input_format = "-f"
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
55 elif (input_format == "fastq") or (input_format == "fastqsanger"):
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
56 input_format = "-q"
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
57 else:
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
58 raise Exception('input format must be one of fasta or fastq')
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
59 if alignment_method == "RNA":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
60 x = "-v %s -M 1 --best --strata -p %s --norc --suppress 6,7,8" % (
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
61 v_mis, pslots)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
62 elif alignment_method == "unique":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
63 x = "-v %s -m 1 -p %s --suppress 6,7,8" % (v_mis, pslots)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
64 elif alignment_method == "multiple":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
65 x = "-v %s -M 1 --best --strata -p %s --suppress 6,7,8" % (
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
66 v_mis, pslots)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
67 elif alignment_method == "k_option":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
68 x = "-v %s -k 1 --best -p %s --suppress 6,7,8" % (v_mis, pslots)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
69 elif alignment_method == "n_option":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
70 x = "-n %s -M 1 --best -p %s --suppress 6,7,8" % (v_mis, pslots)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
71 elif alignment_method == "a_option":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
72 x = "-v %s -a --best -p %s --suppress 6,7,8" % (v_mis, pslots)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
73 if aligned == "None" and unaligned == "None":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
74 fasta_command = ""
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
75 elif aligned != "None" and unaligned == "None":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
76 fasta_command = " --al %s" % aligned
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
77 elif aligned == "None" and unaligned != "None":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
78 fasta_command = " --un %s" % unaligned
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
79 else:
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
80 fasta_command = " --al %s --un %s" % (aligned, unaligned)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
81 x = x + fasta_command
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
82 if out_type == "tabular":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
83 return "bowtie %s %s %s %s > %s" % (x, index, input_format, input, output)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
84 elif out_type == "sam":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
85 return "bowtie %s -S %s %s %s > %s" % (x, index, input_format, input, output)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
86 elif out_type == "bam":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
87 return "bowtie %s -S %s %s %s |samtools view -bS - > %s" % (
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
88 x, index, input_format, input, output)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
89
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
90
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
91 def bowtie_squash(fasta):
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
92 # make temp directory for bowtie indexes
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
93 tmp_index_dir = tempfile.mkdtemp()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
94 ref_file = tempfile.NamedTemporaryFile(dir=tmp_index_dir)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
95 ref_file_name = ref_file.name
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
96 # by default, delete the temporary file, but ref_file.name is now stored
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
97 # in ref_file_name
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
98 ref_file.close()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
99 # symlink between the fasta source file and the deleted ref_file name
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
100 os.symlink(fasta, ref_file_name)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
101 # bowtie command line, which will work after changing dir
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
102 # (cwd=tmp_index_dir)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
103 cmd1 = 'bowtie-build -f %s %s' % (ref_file_name, ref_file_name)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
104 try:
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
105 FNULL = open(os.devnull, 'w')
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
106 # a path string for a temp file in tmp_index_dir. Just a string
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
107 tmp = tempfile.NamedTemporaryFile(dir=tmp_index_dir).name
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
108 # creates and open a file handler pointing to the temp file
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
109 tmp_stderr = open(tmp, 'wb')
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
110 # both stderr and stdout of bowtie-build are redirected in dev/null
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
111 proc = subprocess.Popen(
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
112 args=cmd1, shell=True, cwd=tmp_index_dir, stderr=FNULL, stdout=FNULL)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
113 returncode = proc.wait()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
114 tmp_stderr.close()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
115 FNULL.close()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
116 sys.stdout.write(cmd1 + "\n")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
117 except Exception as e:
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
118 # clean up temp dir
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
119 if os.path.exists(tmp_index_dir):
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
120 shutil.rmtree(tmp_index_dir)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
121 stop_err('Error indexing reference sequence\n' + str(e))
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
122 # no Cleaning if no Exception, tmp_index_dir has to be cleaned after
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
123 # bowtie_alignment()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
124 # bowtie fashion path without extention
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
125 index_full_path = os.path.join(tmp_index_dir, ref_file_name)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
126 return tmp_index_dir, index_full_path
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
127
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
128
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
129 def bowtie_alignment(command_line, flyPreIndexed=''):
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
130 # make temp directory just for stderr
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
131 tmp_index_dir = tempfile.mkdtemp()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
132 tmp = tempfile.NamedTemporaryFile(dir=tmp_index_dir).name
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
133 tmp_stderr = open(tmp, 'wb')
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
134 # conditional statement for sorted bam generation viewable in Trackster
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
135 if "samtools" in command_line:
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
136 # recover the final output file name
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
137 target_file = command_line.split()[-1]
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
138 path_to_unsortedBam = os.path.join(tmp_index_dir, "unsorted.bam")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
139 path_to_sortedBam = os.path.join(tmp_index_dir, "unsorted.bam.sorted")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
140 first_command_line = " ".join(
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
141 command_line.split()[:-3]) + " -o " + path_to_unsortedBam + " - "
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
142 # example: bowtie -v 0 -M 1 --best --strata -p 12 --suppress 6,7,8 -S
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
143 # /home/galaxy/galaxy-dist/bowtie/Dmel/dmel-all-chromosome-r5.49 -f
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
144 # /home/galaxy/galaxy-dist/database/files/003/dataset_3460.dat
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
145 # |samtools view -bS -o /tmp/tmp_PgMT0/unsorted.bam -
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
146 # generates an "unsorted.bam.sorted.bam file", NOT an
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
147 # "unsorted.bam.sorted" file
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
148 second_command_line = "samtools sort %s %s" % (
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
149 path_to_unsortedBam, path_to_sortedBam)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
150 # fileno() method return the file descriptor number of tmp_stderr
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
151 p = subprocess.Popen(
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
152 args=first_command_line, cwd=tmp_index_dir, shell=True, stderr=tmp_stderr.fileno())
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
153 returncode = p.wait()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
154 sys.stdout.write("%s\n" % first_command_line + str(returncode))
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
155 p = subprocess.Popen(
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
156 args=second_command_line, cwd=tmp_index_dir, shell=True, stderr=tmp_stderr.fileno())
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
157 returncode = p.wait()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
158 sys.stdout.write("\n%s\n" % second_command_line + str(returncode))
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
159 if os.path.isfile(path_to_sortedBam + ".bam"):
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
160 shutil.copy2(path_to_sortedBam + ".bam", target_file)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
161 else:
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
162 p = subprocess.Popen(
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
163 args=command_line, shell=True, stderr=tmp_stderr.fileno())
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
164 returncode = p.wait()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
165 sys.stdout.write(command_line + "\n")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
166 tmp_stderr.close()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
167 # cleaning if the index was created in the fly
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
168 if os.path.exists(flyPreIndexed):
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
169 shutil.rmtree(flyPreIndexed)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
170 # cleaning tmp files and directories
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
171 if os.path.exists(tmp_index_dir):
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
172 shutil.rmtree(tmp_index_dir)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
173 return
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
174
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
175
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
176 def __main__():
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
177 args = Parser()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
178 F = open(args.output, "w")
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
179 if args.index_from == "history":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
180 tmp_dir, index_path = bowtie_squash(args.index_source)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
181 else:
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
182 tmp_dir, index_path = "dummy/dymmy", args.index_source
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
183 command_line = bowtieCommandLiner(args.method, args.v_mismatches, args.output_format,
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
184 args.aligned, args.unaligned, args.input_format, args.input,
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
185 index_path, args.output, args.num_threads)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
186 bowtie_alignment(command_line, flyPreIndexed=tmp_dir)
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
187 F.close()
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
188 if __name__ == "__main__":
e8bdae1a2bdc Uploaded
drosofff
parents:
diff changeset
189 __main__()