Mercurial > repos > cstrittmatter > test_galtrakr_eurl_vtec_wgs_pt_23
annotate scripts/rgFastQC.py @ 5:f739b302bc9a draft default tip
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
author | cstrittmatter |
---|---|
date | Wed, 22 Jan 2020 08:12:43 -0500 |
parents | e37910d2c794 |
children |
rev | line source |
---|---|
0
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
1 """ |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
2 Rewrite of rgFastQC.py for Version 0.11.2 of FastQC. |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
3 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
4 Changes implemented from tmcgowan at |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
5 https://testtoolshed.g2.bx.psu.edu/view/tmcgowan/fastqc |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
6 and iuc at https://toolshed.g2.bx.psu.edu/view/iuc/fastqc |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
7 with minor changes and bug fixes |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
8 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
9 SYNOPSIS |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
10 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
11 rgFastQC.py -i input_file -j input_file.name -o output_html_file [-d output_directory] |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
12 [-f fastq|bam|sam] [-n job_name] [-c contaminant_file] [-e fastqc_executable] |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
13 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
14 EXAMPLE (generated by Galaxy) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
15 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
16 rgFastQC.py -i path/dataset_1.dat -j 1000gsample.fastq -o path/dataset_3.dat -d path/job_working_directory/subfolder |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
17 -f fastq -n FastQC -c path/dataset_2.dat -e fastqc |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
18 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
19 """ |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
20 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
21 import re |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
22 import os |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
23 import shutil |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
24 import subprocess |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
25 import optparse |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
26 import tempfile |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
27 import glob |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
28 import gzip |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
29 import bz2 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
30 import zipfile |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
31 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
32 class FastQCRunner(object): |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
33 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
34 def __init__(self,opts=None): |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
35 ''' |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
36 Initializes an object to run FastQC in Galaxy. To start the process, use the function run_fastqc() |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
37 ''' |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
38 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
39 # Check whether the options are specified and saves them into the object |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
40 assert opts != None |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
41 self.opts = opts |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
42 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
43 def prepare_command_line(self): |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
44 ''' |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
45 Develops the Commandline to run FastQC in Galaxy |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
46 ''' |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
47 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
48 # Check whether a given file compression format is valid |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
49 # This prevents uncompression of already uncompressed files |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
50 infname = self.opts.inputfilename |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
51 linf = infname.lower() |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
52 trimext = False |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
53 # decompression at upload currently does NOT remove this now bogus ending - fastqc will barf |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
54 # patched may 29 2013 until this is fixed properly |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
55 if ( linf.endswith('.gz') or linf.endswith('.gzip') ): |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
56 f = gzip.open(self.opts.input) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
57 try: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
58 f.readline() |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
59 except: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
60 trimext = True |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
61 f.close() |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
62 elif linf.endswith('bz2'): |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
63 f = bz2.open(self.opts.input,'rb') |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
64 try: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
65 f.readline() |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
66 except: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
67 trimext = True |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
68 f.close() |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
69 elif linf.endswith('.zip'): |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
70 if not zipfile.is_zipfile(self.opts.input): |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
71 trimext = True |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
72 if trimext: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
73 f = open(self.opts.input) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
74 try: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
75 f.readline() |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
76 except: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
77 raise Exception("Input file corruption, could not identify the filetype") |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
78 infname = os.path.splitext(infname)[0] |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
79 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
80 # Replace unwanted or problematic charaters in the input file name |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
81 self.fastqinfilename = re.sub(ur'[^a-zA-Z0-9_\-\.]', '_', os.path.basename(infname)) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
82 # check that the symbolic link gets a proper ending, fastqc seems to ignore the given format otherwise |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
83 if 'fastq' in opts.informat: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
84 # with fastq the .ext is ignored, but when a format is actually passed it must comply with fastqc's |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
85 # accepted formats.. |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
86 opts.informat = 'fastq' |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
87 elif not self.fastqinfilename.endswith(opts.informat): |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
88 self.fastqinfilename += '.%s' % opts.informat |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
89 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
90 # Build the Commandline from the given parameters |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
91 command_line = [opts.executable, '--outdir %s' % opts.outputdir] |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
92 if opts.contaminants != None: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
93 command_line.append('--contaminants %s' % opts.contaminants) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
94 if opts.limits != None: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
95 command_line.append('--limits %s' % opts.limits) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
96 command_line.append('--quiet') |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
97 command_line.append('--extract') # to access the output text file |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
98 command_line.append(self.fastqinfilename) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
99 command_line.append('-f %s' % opts.informat) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
100 self.command_line = ' '.join(command_line) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
101 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
102 def copy_output_file_to_dataset(self): |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
103 ''' |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
104 Retrieves the output html and text files from the output directory and copies them to the Galaxy output files |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
105 ''' |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
106 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
107 # retrieve html file |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
108 result_file = glob.glob(opts.outputdir + '/*html') |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
109 with open(result_file[0], 'rb') as fsrc: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
110 with open(self.opts.htmloutput, 'wb') as fdest: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
111 shutil.copyfileobj(fsrc, fdest) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
112 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
113 # retrieve text file |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
114 text_file = glob.glob(opts.outputdir + '/*/fastqc_data.txt') |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
115 with open(text_file[0], 'rb') as fsrc: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
116 with open(self.opts.textoutput, 'wb') as fdest: |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
117 shutil.copyfileobj(fsrc, fdest) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
118 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
119 def run_fastqc(self): |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
120 ''' |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
121 Executes FastQC. Make sure the mandatory import parameters input, inputfilename, outputdir and htmloutput have been specified in the options (opts) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
122 ''' |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
123 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
124 # Create a log file |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
125 dummy,tlog = tempfile.mkstemp(prefix='rgFastQC',suffix=".log",dir=self.opts.outputdir) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
126 sout = open(tlog, 'w') |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
127 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
128 self.prepare_command_line() |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
129 sout.write(self.command_line) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
130 sout.write('\n') |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
131 sout.write("Creating symlink\n") # between the input (.dat) file and the given input file name |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
132 os.symlink(self.opts.input, self.fastqinfilename) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
133 sout.write("check_call\n") |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
134 subprocess.check_call(self.command_line, shell=True) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
135 sout.write("Copying working %s file to %s \n" % (self.fastqinfilename, self.opts.htmloutput)) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
136 self.copy_output_file_to_dataset() |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
137 sout.write("Finished") |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
138 sout.close() |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
139 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
140 if __name__ == '__main__': |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
141 op = optparse.OptionParser() |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
142 op.add_option('-i', '--input', default=None) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
143 op.add_option('-j', '--inputfilename', default=None) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
144 op.add_option('-o', '--htmloutput', default=None) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
145 op.add_option('-t', '--textoutput', default=None) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
146 op.add_option('-d', '--outputdir', default="/tmp/shortread") |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
147 op.add_option('-f', '--informat', default='fastq') |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
148 op.add_option('-n', '--namejob', default='rgFastQC') |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
149 op.add_option('-c', '--contaminants', default=None) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
150 op.add_option('-l', '--limits', default=None) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
151 op.add_option('-e', '--executable', default='fastqc') |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
152 opts, args = op.parse_args() |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
153 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
154 assert opts.input != None |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
155 assert opts.inputfilename != None |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
156 assert opts.htmloutput != None |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
157 #assert os.path.isfile(opts.executable),'##rgFastQC.py error - cannot find executable %s' % opts.executable |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
158 if not os.path.exists(opts.outputdir): |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
159 os.makedirs(opts.outputdir) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
160 |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
161 fastqc_runner = FastQCRunner(opts) |
e37910d2c794
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
162 fastqc_runner.run_fastqc() |