Mercurial > repos > devteam > fastqc
comparison rgFastQC.py @ 10:a00a6402d09a draft
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fastqc commit 2bfbb5ae6b801e43355fdc3f964a5111fe3fe3a1
author | iuc |
---|---|
date | Wed, 08 Feb 2017 12:43:43 -0500 |
parents | 3a458e268066 |
children | db2dc6bc8f05 |
comparison
equal
deleted
inserted
replaced
9:3a458e268066 | 10:a00a6402d09a |
---|---|
13 | 13 |
14 EXAMPLE (generated by Galaxy) | 14 EXAMPLE (generated by Galaxy) |
15 | 15 |
16 rgFastQC.py -i path/dataset_1.dat -j 1000gsample.fastq -o path/dataset_3.dat -d path/job_working_directory/subfolder | 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 | 17 -f fastq -n FastQC -c path/dataset_2.dat -e fastqc |
18 | |
19 """ | 18 """ |
20 | 19 import bz2 |
20 import glob | |
21 import gzip | |
22 import mimetypes | |
23 import optparse | |
24 import os | |
21 import re | 25 import re |
22 import os | |
23 import shutil | 26 import shutil |
24 import subprocess | 27 import subprocess |
25 import optparse | |
26 import tempfile | 28 import tempfile |
27 import glob | |
28 import gzip | |
29 import bz2 | |
30 import zipfile | 29 import zipfile |
31 import mimetypes | 30 |
32 | 31 |
33 class FastQCRunner(object): | 32 class FastQCRunner(object): |
34 | 33 def __init__(self, opts=None): |
35 def __init__(self,opts=None): | |
36 ''' | 34 ''' |
37 Initializes an object to run FastQC in Galaxy. To start the process, use the function run_fastqc() | 35 Initializes an object to run FastQC in Galaxy. To start the process, use the function run_fastqc() |
38 ''' | 36 ''' |
39 | 37 |
40 # Check whether the options are specified and saves them into the object | 38 # Check whether the options are specified and saves them into the object |
41 assert opts != None | 39 assert opts is not None |
42 self.opts = opts | 40 self.opts = opts |
43 | 41 |
44 def prepare_command_line(self): | 42 def prepare_command_line(self): |
45 ''' | 43 ''' |
46 Develops the Commandline to run FastQC in Galaxy | 44 Develops the Commandline to run FastQC in Galaxy |
60 f.readline() | 58 f.readline() |
61 except: | 59 except: |
62 trimext = True | 60 trimext = True |
63 f.close() | 61 f.close() |
64 elif linf.endswith('bz2'): | 62 elif linf.endswith('bz2'): |
65 f = bz2.open(self.opts.input,'rb') | 63 f = bz2.BZ2File(self.opts.input, 'r') |
66 try: | 64 try: |
67 f.readline() | 65 f.readline() |
68 except: | 66 except: |
69 trimext = True | 67 trimext = True |
70 f.close() | 68 f.close() |
71 elif linf.endswith('.zip'): | 69 elif linf.endswith('.zip'): |
72 if not zipfile.is_zipfile(self.opts.input): | 70 if not zipfile.is_zipfile(self.opts.input): |
73 trimext = True | 71 trimext = True |
74 if trimext: | 72 if trimext: |
75 f = open(self.opts.input) | 73 f = open(self.opts.input) |
76 try: | 74 try: |
77 f.readline() | 75 f.readline() |
78 except: | 76 except: |
79 raise Exception("Input file corruption, could not identify the filetype") | 77 raise Exception("Input file corruption, could not identify the filetype") |
80 infname = os.path.splitext(infname)[0] | 78 infname = os.path.splitext(infname)[0] |
81 | 79 |
82 # Replace unwanted or problematic charaters in the input file name | 80 # Replace unwanted or problematic charaters in the input file name |
83 self.fastqinfilename = re.sub(ur'[^a-zA-Z0-9_\-\.]', '_', os.path.basename(infname)) | 81 self.fastqinfilename = re.sub(r'[^a-zA-Z0-9_\-\.]', '_', os.path.basename(infname)) |
84 # check that the symbolic link gets a proper ending, fastqc seems to ignore the given format otherwise | 82 # check that the symbolic link gets a proper ending, fastqc seems to ignore the given format otherwise |
85 if 'fastq' in opts.informat: | 83 if 'fastq' in self.opts.informat: |
86 # with fastq the .ext is ignored, but when a format is actually passed it must comply with fastqc's | 84 # with fastq the .ext is ignored, but when a format is actually passed it must comply with fastqc's |
87 # accepted formats.. | 85 # accepted formats.. |
88 opts.informat = 'fastq' | 86 self.opts.informat = 'fastq' |
89 elif not self.fastqinfilename.endswith(opts.informat): | 87 elif not self.fastqinfilename.endswith(self.opts.informat): |
90 self.fastqinfilename += '.%s' % opts.informat | 88 self.fastqinfilename += '.%s' % self.opts.informat |
91 | 89 |
92 # Build the Commandline from the given parameters | 90 # Build the Commandline from the given parameters |
93 command_line = [opts.executable, '--outdir %s' % opts.outputdir] | 91 command_line = [opts.executable, '--outdir %s' % self.opts.outputdir] |
94 if opts.contaminants != None: | 92 if self.opts.contaminants is not None: |
95 command_line.append('--contaminants %s' % opts.contaminants) | 93 command_line.append('--contaminants %s' % self.opts.contaminants) |
96 if opts.limits != None: | 94 if self.opts.limits is not None: |
97 command_line.append('--limits %s' % opts.limits) | 95 command_line.append('--limits %s' % self.opts.limits) |
98 command_line.append('--quiet') | 96 command_line.append('--quiet') |
99 command_line.append('--extract') # to access the output text file | 97 command_line.append('--extract') # to access the output text file |
100 if type[-1] != "gzip": | 98 if type[-1] != "gzip": |
101 command_line.append('-f %s' % opts.informat) | 99 command_line.append('-f %s' % self.opts.informat) |
102 else: | 100 else: |
103 self.fastqinfilename += ".gz" | 101 self.fastqinfilename += ".gz" |
104 command_line.append(self.fastqinfilename) | 102 command_line.append(self.fastqinfilename) |
105 self.command_line = ' '.join(command_line) | 103 self.command_line = ' '.join(command_line) |
106 | 104 |
107 def copy_output_file_to_dataset(self): | 105 def copy_output_file_to_dataset(self): |
108 ''' | 106 ''' |
109 Retrieves the output html and text files from the output directory and copies them to the Galaxy output files | 107 Retrieves the output html and text files from the output directory and copies them to the Galaxy output files |
110 ''' | 108 ''' |
111 | 109 |
112 # retrieve html file | 110 # retrieve html file |
113 result_file = glob.glob(opts.outputdir + '/*html') | 111 result_file = glob.glob(self.opts.outputdir + '/*html') |
114 with open(result_file[0], 'rb') as fsrc: | 112 with open(result_file[0], 'rb') as fsrc: |
115 with open(self.opts.htmloutput, 'wb') as fdest: | 113 with open(self.opts.htmloutput, 'wb') as fdest: |
116 shutil.copyfileobj(fsrc, fdest) | 114 shutil.copyfileobj(fsrc, fdest) |
117 | 115 |
118 # retrieve text file | 116 # retrieve text file |
119 text_file = glob.glob(opts.outputdir + '/*/fastqc_data.txt') | 117 text_file = glob.glob(self.opts.outputdir + '/*/fastqc_data.txt') |
120 with open(text_file[0], 'rb') as fsrc: | 118 with open(text_file[0], 'rb') as fsrc: |
121 with open(self.opts.textoutput, 'wb') as fdest: | 119 with open(self.opts.textoutput, 'wb') as fdest: |
122 shutil.copyfileobj(fsrc, fdest) | 120 shutil.copyfileobj(fsrc, fdest) |
123 | 121 |
124 def run_fastqc(self): | 122 def run_fastqc(self): |
125 ''' | 123 ''' |
126 Executes FastQC. Make sure the mandatory import parameters input, inputfilename, outputdir and htmloutput have been specified in the options (opts) | 124 Executes FastQC. Make sure the mandatory import parameters input, inputfilename, outputdir and htmloutput have been specified in the options |
127 ''' | 125 ''' |
128 | 126 |
129 # Create a log file | 127 # Create a log file |
130 dummy,tlog = tempfile.mkstemp(prefix='rgFastQC',suffix=".log",dir=self.opts.outputdir) | 128 dummy, tlog = tempfile.mkstemp(prefix='rgFastQC', suffix=".log", dir=self.opts.outputdir) |
131 sout = open(tlog, 'w') | 129 sout = open(tlog, 'w') |
132 | 130 |
133 self.prepare_command_line() | 131 self.prepare_command_line() |
134 sout.write(self.command_line) | 132 sout.write(self.command_line) |
135 sout.write('\n') | 133 sout.write('\n') |
136 sout.write("Creating symlink\n") # between the input (.dat) file and the given input file name | 134 sout.write("Creating symlink\n") # between the input (.dat) file and the given input file name |
137 os.symlink(self.opts.input, self.fastqinfilename) | 135 os.symlink(self.opts.input, self.fastqinfilename) |
138 sout.write("check_call\n") | 136 sout.write("check_call\n") |
139 subprocess.check_call(self.command_line, shell=True) | 137 subprocess.check_call(self.command_line, shell=True) |
140 sout.write("Copying working %s file to %s \n" % (self.fastqinfilename, self.opts.htmloutput)) | 138 sout.write("Copying working %s file to %s \n" % (self.fastqinfilename, self.opts.htmloutput)) |
141 self.copy_output_file_to_dataset() | 139 self.copy_output_file_to_dataset() |
142 sout.write("Finished") | 140 sout.write("Finished") |
143 sout.close() | 141 sout.close() |
142 | |
144 | 143 |
145 if __name__ == '__main__': | 144 if __name__ == '__main__': |
146 op = optparse.OptionParser() | 145 op = optparse.OptionParser() |
147 op.add_option('-i', '--input', default=None) | 146 op.add_option('-i', '--input', default=None) |
148 op.add_option('-j', '--inputfilename', default=None) | 147 op.add_option('-j', '--inputfilename', default=None) |
154 op.add_option('-c', '--contaminants', default=None) | 153 op.add_option('-c', '--contaminants', default=None) |
155 op.add_option('-l', '--limits', default=None) | 154 op.add_option('-l', '--limits', default=None) |
156 op.add_option('-e', '--executable', default='fastqc') | 155 op.add_option('-e', '--executable', default='fastqc') |
157 opts, args = op.parse_args() | 156 opts, args = op.parse_args() |
158 | 157 |
159 assert opts.input != None | 158 assert opts.input is not None |
160 assert opts.inputfilename != None | 159 assert opts.inputfilename is not None |
161 assert opts.htmloutput != None | 160 assert opts.htmloutput is not None |
162 if not os.path.exists(opts.outputdir): | 161 if not os.path.exists(opts.outputdir): |
163 os.makedirs(opts.outputdir) | 162 os.makedirs(opts.outputdir) |
164 | 163 |
165 fastqc_runner = FastQCRunner(opts) | 164 fastqc_runner = FastQCRunner(opts) |
166 fastqc_runner.run_fastqc() | 165 fastqc_runner.run_fastqc() |