|
0
|
1 """
|
|
1
|
2 Rewrite of rgFastQC.py for Version 0.11.2 of FastQC.
|
|
|
3
|
|
|
4 Changes implemented from tmcgowan at
|
|
|
5 https://testtoolshed.g2.bx.psu.edu/view/tmcgowan/fastqc
|
|
|
6 and iuc at https://toolshed.g2.bx.psu.edu/view/iuc/fastqc
|
|
|
7 with minor changes and bug fixes
|
|
0
|
8
|
|
1
|
9 SYNOPSIS
|
|
|
10
|
|
|
11 rgFastQC.py -i input_file -j input_file.name -o output_html_file [-d output_directory]
|
|
|
12 [-f fastq|bam|sam] [-n job_name] [-c contaminant_file] [-e fastqc_executable]
|
|
0
|
13
|
|
1
|
14 EXAMPLE (generated by Galaxy)
|
|
0
|
15
|
|
1
|
16 rgFastQC.py -i path/dataset_1.dat -j 1000gsample.fastq -o path/dataset_3.dat -d path/job_working_directory/subfolder
|
|
|
17 -f fastq -n FastQC -c path/dataset_2.dat -e fastqc
|
|
0
|
18
|
|
|
19 """
|
|
1
|
20
|
|
0
|
21 import re
|
|
|
22 import os
|
|
1
|
23 import shutil
|
|
0
|
24 import subprocess
|
|
|
25 import optparse
|
|
|
26 import tempfile
|
|
1
|
27 import glob
|
|
|
28 import gzip
|
|
|
29 import bz2
|
|
0
|
30 import zipfile
|
|
|
31
|
|
1
|
32 class FastQCRunner(object):
|
|
0
|
33
|
|
|
34 def __init__(self,opts=None):
|
|
1
|
35 '''
|
|
|
36 Initializes an object to run FastQC in Galaxy. To start the process, use the function run_fastqc()
|
|
|
37 '''
|
|
0
|
38
|
|
1
|
39 # Check whether the options are specified and saves them into the object
|
|
|
40 assert opts != None
|
|
|
41 self.opts = opts
|
|
0
|
42
|
|
1
|
43 def prepare_command_line(self):
|
|
|
44 '''
|
|
|
45 Develops the Commandline to run FastQC in Galaxy
|
|
|
46 '''
|
|
|
47
|
|
|
48 # Check whether a given file compression format is valid
|
|
|
49 # This prevents uncompression of already uncompressed files
|
|
0
|
50 infname = self.opts.inputfilename
|
|
|
51 linf = infname.lower()
|
|
|
52 trimext = False
|
|
|
53 # decompression at upload currently does NOT remove this now bogus ending - fastqc will barf
|
|
|
54 # patched may 29 2013 until this is fixed properly
|
|
|
55 if ( linf.endswith('.gz') or linf.endswith('.gzip') ):
|
|
|
56 f = gzip.open(self.opts.input)
|
|
|
57 try:
|
|
1
|
58 f.readline()
|
|
0
|
59 except:
|
|
|
60 trimext = True
|
|
|
61 f.close()
|
|
|
62 elif linf.endswith('bz2'):
|
|
|
63 f = bz2.open(self.opts.input,'rb')
|
|
|
64 try:
|
|
|
65 f.readline()
|
|
|
66 except:
|
|
|
67 trimext = True
|
|
|
68 f.close()
|
|
|
69 elif linf.endswith('.zip'):
|
|
|
70 if not zipfile.is_zipfile(self.opts.input):
|
|
|
71 trimext = True
|
|
|
72 if trimext:
|
|
1
|
73 f = open(self.opts.input)
|
|
|
74 try:
|
|
|
75 f.readline()
|
|
|
76 except:
|
|
|
77 raise Exception("Input file corruption, could not identify the filetype")
|
|
|
78 infname = os.path.splitext(infname)[0]
|
|
|
79
|
|
|
80 # Replace unwanted or problematic charaters in the input file name
|
|
|
81 self.fastqinfilename = re.sub(ur'[^a-zA-Z0-9_\-\.]', '_', os.path.basename(infname))
|
|
4
|
82 # check that the symbolic link gets a proper ending, fastqc seems to ignore the given format otherwise
|
|
|
83 if 'fastq' in opts.informat:
|
|
|
84 # with fastq the .ext is ignored, but when a format is actually passed it must comply with fastqc's
|
|
|
85 # accepted formats..
|
|
|
86 opts.informat = 'fastq'
|
|
|
87 elif not self.fastqinfilename.endswith(opts.informat):
|
|
|
88 self.fastqinfilename += '.%s' % opts.informat
|
|
|
89
|
|
1
|
90 # Build the Commandline from the given parameters
|
|
|
91 command_line = [opts.executable, '--outdir %s' % opts.outputdir]
|
|
|
92 if opts.contaminants != None:
|
|
|
93 command_line.append('--contaminants %s' % opts.contaminants)
|
|
|
94 if opts.limits != None:
|
|
|
95 command_line.append('--limits %s' % opts.limits)
|
|
|
96 command_line.append('--quiet')
|
|
|
97 command_line.append('--extract') # to access the output text file
|
|
|
98 command_line.append(self.fastqinfilename)
|
|
4
|
99 command_line.append('-f %s' % opts.informat)
|
|
1
|
100 self.command_line = ' '.join(command_line)
|
|
|
101
|
|
|
102 def copy_output_file_to_dataset(self):
|
|
|
103 '''
|
|
|
104 Retrieves the output html and text files from the output directory and copies them to the Galaxy output files
|
|
|
105 '''
|
|
|
106
|
|
|
107 # retrieve html file
|
|
|
108 result_file = glob.glob(opts.outputdir + '/*html')
|
|
|
109 with open(result_file[0], 'rb') as fsrc:
|
|
|
110 with open(self.opts.htmloutput, 'wb') as fdest:
|
|
|
111 shutil.copyfileobj(fsrc, fdest)
|
|
|
112
|
|
|
113 # retrieve text file
|
|
|
114 text_file = glob.glob(opts.outputdir + '/*/fastqc_data.txt')
|
|
|
115 with open(text_file[0], 'rb') as fsrc:
|
|
|
116 with open(self.opts.textoutput, 'wb') as fdest:
|
|
|
117 shutil.copyfileobj(fsrc, fdest)
|
|
|
118
|
|
|
119 def run_fastqc(self):
|
|
|
120 '''
|
|
|
121 Executes FastQC. Make sure the mandatory import parameters input, inputfilename, outputdir and htmloutput have been specified in the options (opts)
|
|
|
122 '''
|
|
|
123
|
|
|
124 # Create a log file
|
|
|
125 dummy,tlog = tempfile.mkstemp(prefix='rgFastQC',suffix=".log",dir=self.opts.outputdir)
|
|
|
126 sout = open(tlog, 'w')
|
|
|
127
|
|
|
128 self.prepare_command_line()
|
|
|
129 sout.write(self.command_line)
|
|
|
130 sout.write('\n')
|
|
|
131 sout.write("Creating symlink\n") # between the input (.dat) file and the given input file name
|
|
|
132 os.symlink(self.opts.input, self.fastqinfilename)
|
|
|
133 sout.write("check_call\n")
|
|
|
134 subprocess.check_call(self.command_line, shell=True)
|
|
|
135 sout.write("Copying working %s file to %s \n" % (self.fastqinfilename, self.opts.htmloutput))
|
|
|
136 self.copy_output_file_to_dataset()
|
|
|
137 sout.write("Finished")
|
|
0
|
138 sout.close()
|
|
|
139
|
|
|
140 if __name__ == '__main__':
|
|
|
141 op = optparse.OptionParser()
|
|
|
142 op.add_option('-i', '--input', default=None)
|
|
1
|
143 op.add_option('-j', '--inputfilename', default=None)
|
|
0
|
144 op.add_option('-o', '--htmloutput', default=None)
|
|
1
|
145 op.add_option('-t', '--textoutput', default=None)
|
|
0
|
146 op.add_option('-d', '--outputdir', default="/tmp/shortread")
|
|
|
147 op.add_option('-f', '--informat', default='fastq')
|
|
|
148 op.add_option('-n', '--namejob', default='rgFastQC')
|
|
|
149 op.add_option('-c', '--contaminants', default=None)
|
|
1
|
150 op.add_option('-l', '--limits', default=None)
|
|
0
|
151 op.add_option('-e', '--executable', default='fastqc')
|
|
|
152 opts, args = op.parse_args()
|
|
1
|
153
|
|
|
154 assert opts.input != None
|
|
|
155 assert opts.inputfilename != None
|
|
|
156 assert opts.htmloutput != None
|
|
0
|
157 assert os.path.isfile(opts.executable),'##rgFastQC.py error - cannot find executable %s' % opts.executable
|
|
|
158 if not os.path.exists(opts.outputdir):
|
|
|
159 os.makedirs(opts.outputdir)
|
|
|
160
|
|
1
|
161 fastqc_runner = FastQCRunner(opts)
|
|
3
|
162 fastqc_runner.run_fastqc()
|
