Mercurial > repos > peterjc > mira4_assembler

comparison tools/mira4_0/mira4_convert.py @ 32:56b421d59805 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
author peterjc
date Thu, 18 May 2017 13:36:08 -0400
parents fd95aaef8818
children 0785a6537f3e
comparison
equal deleted inserted replaced
31:fd95aaef8818 32:56b421d59805
1 #!/usr/bin/env python 1 #!/usr/bin/env python
2 """A simple wrapper script to call MIRA and collect its output. 2 """A simple wrapper script to call MIRA and collect its output.
3 3
4 This focuses on the miraconvert binary. 4 This focuses on the miraconvert binary.
5 """ 5 """
6
7 from __future__ import print_function
8
6 import os 9 import os
10 import shutil
11 import subprocess
7 import sys 12 import sys
8 import subprocess 13
9 import shutil
10 from optparse import OptionParser 14 from optparse import OptionParser
15
11 try: 16 try:
12 from io import BytesIO 17 from io import BytesIO
13 except ImportError: 18 except ImportError:
14 #Should we worry about Python 2.5 or older? 19 # Should we worry about Python 2.5 or older?
15 from StringIO import StringIO as BytesIO 20 from StringIO import StringIO as BytesIO
16 21
17 #Do we need any PYTHONPATH magic? 22 # Do we need any PYTHONPATH magic?
18 from mira4_make_bam import depad 23 from mira4_make_bam import depad
19 24
20 WRAPPER_VER = "0.0.7" # Keep in sync with the XML file 25 WRAPPER_VER = "0.0.10" # Keep in sync with the XML file
21 26
22 27
23 def run(cmd): 28 def run(cmd):
24 #Avoid using shell=True when we call subprocess to ensure if the Python 29 # Avoid using shell=True when we call subprocess to ensure if the Python
25 #script is killed, so too is the child process. 30 # script is killed, so too is the child process.
26 try: 31 try:
27 child = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE) 32 child = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
28 except Exception, err: 33 except Exception as err:
29 sys.exit("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err)) 34 sys.exit("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
30 #Use .communicate as can get deadlocks with .wait(), 35 # Use .communicate as can get deadlocks with .wait(),
31 stdout, stderr = child.communicate() 36 stdout, stderr = child.communicate()
32 return_code = child.returncode 37 return_code = child.returncode
33 if return_code: 38 if return_code:
34 cmd_str = " ".join(cmd) # doesn't quote spaces etc 39 cmd_str = " ".join(cmd) # doesn't quote spaces etc
35 if stderr and stdout: 40 if stderr and stdout:
36 sys.exit("Return code %i from command:\n%s\n\n%s\n\n%s" % (return_code, cmd_str, stdout, stderr)) 41 sys.exit("Return code %i from command:\n%s\n\n%s\n\n%s" % (return_code, cmd_str, stdout, stderr))
37 else: 42 else:
38 sys.exit("Return code %i from command:\n%s\n%s" % (return_code, cmd_str, stderr)) 43 sys.exit("Return code %i from command:\n%s\n%s" % (return_code, cmd_str, stderr))
44
39 45
40 def get_version(mira_binary): 46 def get_version(mira_binary):
41 """Run MIRA to find its version number""" 47 """Run MIRA to find its version number"""
42 # At the commend line I would use: mira -v | head -n 1 48 # At the commend line I would use: mira -v | head -n 1
43 # however there is some pipe error when doing that here. 49 # however there is some pipe error when doing that here.
44 cmd = [mira_binary, "-v"] 50 cmd = [mira_binary, "-v"]
45 try: 51 try:
46 child = subprocess.Popen(cmd, 52 child = subprocess.Popen(cmd,
47 stdout=subprocess.PIPE, 53 stdout=subprocess.PIPE,
48 stderr=subprocess.STDOUT) 54 stderr=subprocess.STDOUT)
49 except Exception, err: 55 except Exception as err:
50 sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err)) 56 sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
51 sys.exit(1) 57 sys.exit(1)
52 ver, tmp = child.communicate() 58 ver, tmp = child.communicate()
53 del child 59 del child
54 return ver.split("\n", 1)[0].strip() 60 return ver.split("\n", 1)[0].strip()
55 61
56 #Parse Command Line 62
63 # Parse Command Line
57 usage = """Galaxy MIRA4 wrapper script v%s - use as follows: 64 usage = """Galaxy MIRA4 wrapper script v%s - use as follows:
58 65
59 $ python mira4_convert.py ... 66 $ python mira4_convert.py ...
60 67
61 This will run the MIRA miraconvert binary and collect its output files as directed. 68 This will run the MIRA miraconvert binary and collect its output files as directed.
134 sys.exit("Bad %s setting, %r" % (name, value)) 141 sys.exit("Bad %s setting, %r" % (name, value))
135 if i < 0: 142 if i < 0:
136 sys.exit("Negative %s setting, %r" % (name, value)) 143 sys.exit("Negative %s setting, %r" % (name, value))
137 return i 144 return i
138 145
146
139 min_length = check_min_int(options.min_length, "minimum length") 147 min_length = check_min_int(options.min_length, "minimum length")
140 min_cover = check_min_int(options.min_cover, "minimum cover") 148 min_cover = check_min_int(options.min_cover, "minimum cover")
141 min_reads = check_min_int(options.min_reads, "minimum reads") 149 min_reads = check_min_int(options.min_reads, "minimum reads")
142 150
143 #TODO - Run MIRA in /tmp or a configurable directory? 151 # TODO - Run MIRA in /tmp or a configurable directory?
144 #Currently Galaxy puts us somewhere safe like: 152 # Currently Galaxy puts us somewhere safe like:
145 #/opt/galaxy-dist/database/job_working_directory/846/ 153 # /opt/galaxy-dist/database/job_working_directory/846/
146 temp = "." 154 temp = "."
147 155
148 156
149 cmd_list = [mira_convert] 157 cmd_list = [mira_convert]
150 if min_length: 158 if min_length:
157 if out_maf: 165 if out_maf:
158 cmd_list.append("maf") 166 cmd_list.append("maf")
159 if out_bam: 167 if out_bam:
160 cmd_list.append("samnbb") 168 cmd_list.append("samnbb")
161 if not out_fasta: 169 if not out_fasta:
162 #Need this for samtools depad 170 # Need this for samtools depad
163 out_fasta = os.path.join(temp, "depadded.fasta") 171 out_fasta = os.path.join(temp, "depadded.fasta")
164 if out_fasta: 172 if out_fasta:
165 cmd_list.append("fasta") 173 cmd_list.append("fasta")
166 if out_ace: 174 if out_ace:
167 cmd_list.append("ace") 175 cmd_list.append("ace")
168 if out_cstats: 176 if out_cstats:
169 cmd_list.append("cstats") 177 cmd_list.append("cstats")
170 run(cmd_list) 178 run(cmd_list)
171 179
180
172 def collect(old, new): 181 def collect(old, new):
173 if not os.path.isfile(old): 182 if not os.path.isfile(old):
174 sys.exit("Missing expected output file %s" % old) 183 sys.exit("Missing expected output file %s" % old)
175 shutil.move(old, new) 184 shutil.move(old, new)
176 185
186
177 if out_maf: 187 if out_maf:
178 collect(os.path.join(temp, "converted.maf"), out_maf) 188 collect(os.path.join(temp, "converted.maf"), out_maf)
179 if out_fasta: 189 if out_fasta:
180 #Can we look at the MAF file to see if there are multiple strains? 190 # Can we look at the MAF file to see if there are multiple strains?
181 old = os.path.join(temp, "converted_AllStrains.unpadded.fasta") 191 old = os.path.join(temp, "converted_AllStrains.unpadded.fasta")
182 if os.path.isfile(old): 192 if os.path.isfile(old):
183 collect(old, out_fasta) 193 collect(old, out_fasta)
184 else: 194 else:
185 #Might the output be filtered down to zero contigs? 195 # Might the output be filtered down to zero contigs?
186 old = os.path.join(temp, "converted.fasta") 196 old = os.path.join(temp, "converted.fasta")
187 if not os.path.isfile(old): 197 if not os.path.isfile(old):
188 sys.exit("Missing expected output FASTA file") 198 sys.exit("Missing expected output FASTA file")
189 elif os.path.getsize(old) == 0: 199 elif os.path.getsize(old) == 0:
190 print("Warning - no contigs (harsh filters?)") 200 print("Warning - no contigs (harsh filters?)")
210 print(h.getvalue()) 220 print(h.getvalue())
211 h.close() 221 h.close()
212 sys.exit(1) 222 sys.exit(1)
213 h.close() 223 h.close()
214 if out_fasta == os.path.join(temp, "depadded.fasta"): 224 if out_fasta == os.path.join(temp, "depadded.fasta"):
215 #Not asked for by Galaxy, no longer needed 225 # Not asked for by Galaxy, no longer needed
216 os.remove(out_fasta) 226 os.remove(out_fasta)
217 227
218 if min_length or min_cover or min_reads: 228 if min_length or min_cover or min_reads:
219 print("Filtered.") 229 print("Filtered.")
220 else: 230 else: