Mercurial > repos > devteam > fastqc
annotate rgFastQC.py @ 9:3a458e268066 draft
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit 8918618a5ef7bdca55a31cd919efa593044a376e
| author | devteam |
|---|---|
| date | Wed, 02 Nov 2016 16:12:51 -0400 |
| parents | 06819360a9e2 |
| children | a00a6402d09a |
| rev | line source |
|---|---|
| 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 |
|
9
3a458e268066
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit 8918618a5ef7bdca55a31cd919efa593044a376e
devteam
parents:
8
diff
changeset
|
31 import mimetypes |
| 0 | 32 |
| 1 | 33 class FastQCRunner(object): |
|
8
06819360a9e2
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit df4c0b0c6372e2984966e220fa42ecd8a3d370e8
devteam
parents:
7
diff
changeset
|
34 |
| 0 | 35 def __init__(self,opts=None): |
| 1 | 36 ''' |
| 37 Initializes an object to run FastQC in Galaxy. To start the process, use the function run_fastqc() | |
| 38 ''' | |
|
8
06819360a9e2
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit df4c0b0c6372e2984966e220fa42ecd8a3d370e8
devteam
parents:
7
diff
changeset
|
39 |
| 1 | 40 # Check whether the options are specified and saves them into the object |
| 41 assert opts != None | |
| 42 self.opts = opts | |
| 0 | 43 |
| 1 | 44 def prepare_command_line(self): |
| 45 ''' | |
| 46 Develops the Commandline to run FastQC in Galaxy | |
| 47 ''' | |
|
8
06819360a9e2
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit df4c0b0c6372e2984966e220fa42ecd8a3d370e8
devteam
parents:
7
diff
changeset
|
48 |
| 1 | 49 # Check whether a given file compression format is valid |
| 50 # This prevents uncompression of already uncompressed files | |
| 0 | 51 infname = self.opts.inputfilename |
| 52 linf = infname.lower() | |
| 53 trimext = False | |
| 54 # decompression at upload currently does NOT remove this now bogus ending - fastqc will barf | |
| 55 # patched may 29 2013 until this is fixed properly | |
|
9
3a458e268066
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit 8918618a5ef7bdca55a31cd919efa593044a376e
devteam
parents:
8
diff
changeset
|
56 type = mimetypes.guess_type(self.opts.input) |
|
3a458e268066
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit 8918618a5ef7bdca55a31cd919efa593044a376e
devteam
parents:
8
diff
changeset
|
57 if linf.endswith('.gz') or linf.endswith('.gzip') or type[-1] == "gzip": |
| 0 | 58 f = gzip.open(self.opts.input) |
| 59 try: | |
| 1 | 60 f.readline() |
| 0 | 61 except: |
| 62 trimext = True | |
| 63 f.close() | |
| 64 elif linf.endswith('bz2'): | |
| 65 f = bz2.open(self.opts.input,'rb') | |
| 66 try: | |
| 67 f.readline() | |
| 68 except: | |
| 69 trimext = True | |
| 70 f.close() | |
| 71 elif linf.endswith('.zip'): | |
| 72 if not zipfile.is_zipfile(self.opts.input): | |
| 73 trimext = True | |
| 74 if trimext: | |
| 1 | 75 f = open(self.opts.input) |
| 76 try: | |
| 77 f.readline() | |
| 78 except: | |
| 79 raise Exception("Input file corruption, could not identify the filetype") | |
| 80 infname = os.path.splitext(infname)[0] | |
|
8
06819360a9e2
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit df4c0b0c6372e2984966e220fa42ecd8a3d370e8
devteam
parents:
7
diff
changeset
|
81 |
| 1 | 82 # Replace unwanted or problematic charaters in the input file name |
| 83 self.fastqinfilename = re.sub(ur'[^a-zA-Z0-9_\-\.]', '_', os.path.basename(infname)) | |
| 4 | 84 # check that the symbolic link gets a proper ending, fastqc seems to ignore the given format otherwise |
| 85 if 'fastq' in opts.informat: | |
|
8
06819360a9e2
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit df4c0b0c6372e2984966e220fa42ecd8a3d370e8
devteam
parents:
7
diff
changeset
|
86 # with fastq the .ext is ignored, but when a format is actually passed it must comply with fastqc's |
| 4 | 87 # accepted formats.. |
| 88 opts.informat = 'fastq' | |
| 89 elif not self.fastqinfilename.endswith(opts.informat): | |
| 90 self.fastqinfilename += '.%s' % opts.informat | |
| 91 | |
| 1 | 92 # Build the Commandline from the given parameters |
| 93 command_line = [opts.executable, '--outdir %s' % opts.outputdir] | |
| 94 if opts.contaminants != None: | |
| 95 command_line.append('--contaminants %s' % opts.contaminants) | |
| 96 if opts.limits != None: | |
| 97 command_line.append('--limits %s' % opts.limits) | |
| 98 command_line.append('--quiet') | |
| 99 command_line.append('--extract') # to access the output text file | |
|
9
3a458e268066
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit 8918618a5ef7bdca55a31cd919efa593044a376e
devteam
parents:
8
diff
changeset
|
100 if type[-1] != "gzip": |
|
3a458e268066
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit 8918618a5ef7bdca55a31cd919efa593044a376e
devteam
parents:
8
diff
changeset
|
101 command_line.append('-f %s' % opts.informat) |
|
3a458e268066
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit 8918618a5ef7bdca55a31cd919efa593044a376e
devteam
parents:
8
diff
changeset
|
102 else: |
|
3a458e268066
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit 8918618a5ef7bdca55a31cd919efa593044a376e
devteam
parents:
8
diff
changeset
|
103 self.fastqinfilename += ".gz" |
| 1 | 104 command_line.append(self.fastqinfilename) |
| 105 self.command_line = ' '.join(command_line) | |
| 106 | |
| 107 def copy_output_file_to_dataset(self): | |
| 108 ''' | |
| 109 Retrieves the output html and text files from the output directory and copies them to the Galaxy output files | |
| 110 ''' | |
|
8
06819360a9e2
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit df4c0b0c6372e2984966e220fa42ecd8a3d370e8
devteam
parents:
7
diff
changeset
|
111 |
| 1 | 112 # retrieve html file |
| 113 result_file = glob.glob(opts.outputdir + '/*html') | |
| 114 with open(result_file[0], 'rb') as fsrc: | |
| 115 with open(self.opts.htmloutput, 'wb') as fdest: | |
| 116 shutil.copyfileobj(fsrc, fdest) | |
|
8
06819360a9e2
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit df4c0b0c6372e2984966e220fa42ecd8a3d370e8
devteam
parents:
7
diff
changeset
|
117 |
| 1 | 118 # retrieve text file |
| 119 text_file = glob.glob(opts.outputdir + '/*/fastqc_data.txt') | |
| 120 with open(text_file[0], 'rb') as fsrc: | |
| 121 with open(self.opts.textoutput, 'wb') as fdest: | |
| 122 shutil.copyfileobj(fsrc, fdest) | |
| 123 | |
| 124 def run_fastqc(self): | |
| 125 ''' | |
| 126 Executes FastQC. Make sure the mandatory import parameters input, inputfilename, outputdir and htmloutput have been specified in the options (opts) | |
| 127 ''' | |
|
8
06819360a9e2
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit df4c0b0c6372e2984966e220fa42ecd8a3d370e8
devteam
parents:
7
diff
changeset
|
128 |
| 1 | 129 # Create a log file |
| 130 dummy,tlog = tempfile.mkstemp(prefix='rgFastQC',suffix=".log",dir=self.opts.outputdir) | |
| 131 sout = open(tlog, 'w') | |
|
8
06819360a9e2
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit df4c0b0c6372e2984966e220fa42ecd8a3d370e8
devteam
parents:
7
diff
changeset
|
132 |
| 1 | 133 self.prepare_command_line() |
| 134 sout.write(self.command_line) | |
| 135 sout.write('\n') | |
| 136 sout.write("Creating symlink\n") # between the input (.dat) file and the given input file name | |
| 137 os.symlink(self.opts.input, self.fastqinfilename) | |
| 138 sout.write("check_call\n") | |
| 139 subprocess.check_call(self.command_line, shell=True) | |
| 140 sout.write("Copying working %s file to %s \n" % (self.fastqinfilename, self.opts.htmloutput)) | |
| 141 self.copy_output_file_to_dataset() | |
| 142 sout.write("Finished") | |
| 0 | 143 sout.close() |
| 144 | |
| 145 if __name__ == '__main__': | |
| 146 op = optparse.OptionParser() | |
| 147 op.add_option('-i', '--input', default=None) | |
| 1 | 148 op.add_option('-j', '--inputfilename', default=None) |
| 0 | 149 op.add_option('-o', '--htmloutput', default=None) |
| 1 | 150 op.add_option('-t', '--textoutput', default=None) |
| 0 | 151 op.add_option('-d', '--outputdir', default="/tmp/shortread") |
| 152 op.add_option('-f', '--informat', default='fastq') | |
| 153 op.add_option('-n', '--namejob', default='rgFastQC') | |
| 154 op.add_option('-c', '--contaminants', default=None) | |
| 1 | 155 op.add_option('-l', '--limits', default=None) |
| 0 | 156 op.add_option('-e', '--executable', default='fastqc') |
| 157 opts, args = op.parse_args() | |
|
8
06819360a9e2
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit df4c0b0c6372e2984966e220fa42ecd8a3d370e8
devteam
parents:
7
diff
changeset
|
158 |
| 1 | 159 assert opts.input != None |
| 160 assert opts.inputfilename != None | |
| 161 assert opts.htmloutput != None | |
|
8
06819360a9e2
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit df4c0b0c6372e2984966e220fa42ecd8a3d370e8
devteam
parents:
7
diff
changeset
|
162 if not os.path.exists(opts.outputdir): |
| 0 | 163 os.makedirs(opts.outputdir) |
|
8
06819360a9e2
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fastqc commit df4c0b0c6372e2984966e220fa42ecd8a3d370e8
devteam
parents:
7
diff
changeset
|
164 |
| 1 | 165 fastqc_runner = FastQCRunner(opts) |
| 3 | 166 fastqc_runner.run_fastqc() |