|
0
|
1 #Dan Blankenberg
|
|
|
2 import sys, os, shutil
|
|
|
3 import imp
|
|
|
4 from galaxy_utils.sequence.fastq import fastqReader, fastqWriter
|
|
|
5
|
|
|
6 def main():
|
|
|
7 #Read command line arguments
|
|
|
8 input_filename = sys.argv[1]
|
|
|
9 script_filename = sys.argv[2]
|
|
|
10 output_filename = sys.argv[3]
|
|
|
11 additional_files_path = sys.argv[4]
|
|
|
12 input_type = sys.argv[5] or 'sanger'
|
|
|
13
|
|
|
14 #Save script file for debuging/verification info later
|
|
|
15 os.mkdir( additional_files_path )
|
|
|
16 shutil.copy( script_filename, os.path.join( additional_files_path, 'debug.txt' ) )
|
|
|
17
|
|
|
18 fastq_manipulator = imp.load_module( 'fastq_manipulator', open( script_filename ), script_filename, ( '', 'r', imp.PY_SOURCE ) )
|
|
|
19
|
|
|
20 out = fastqWriter( open( output_filename, 'wb' ), format = input_type )
|
|
|
21
|
|
|
22 i = None
|
|
|
23 reads_manipulated = 0
|
|
|
24 for i, fastq_read in enumerate( fastqReader( open( input_filename ), format = input_type ) ):
|
|
|
25 new_read = fastq_manipulator.match_and_manipulate_read( fastq_read )
|
|
|
26 if new_read:
|
|
|
27 out.write( new_read )
|
|
|
28 if new_read != fastq_read:
|
|
|
29 reads_manipulated += 1
|
|
|
30 out.close()
|
|
|
31 if i is None:
|
|
|
32 print "Your file contains no valid FASTQ reads."
|
|
|
33 else:
|
|
|
34 print 'Manipulated %s of %s reads (%.2f%%).' % ( reads_manipulated, i + 1, float( reads_manipulated ) / float( i + 1 ) * 100.0 )
|
|
|
35
|
|
|
36 if __name__ == "__main__":
|
|
|
37 main()
|
