comparison fastq_paired_end_interlacer.py @ 0:b89bdf6acb6c draft

Imported from capsule None
author devteam
date Mon, 27 Jan 2014 09:26:38 -0500
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:b89bdf6acb6c
1 #Florent Angly
2 import sys
3 from galaxy_utils.sequence.fastq import fastqReader, fastqWriter, fastqNamedReader, fastqJoiner
4
5 def main():
6 mate1_filename = sys.argv[1]
7 mate1_type = sys.argv[2] or 'sanger'
8 mate2_filename = sys.argv[3]
9 mate2_type = sys.argv[4] or 'sanger'
10 outfile_pairs = sys.argv[5]
11 outfile_singles = sys.argv[6]
12
13 if mate1_type != mate2_type:
14 print "WARNING: You are trying to interlace files of two different types: %s and %s." % ( mate1_type, mate2_type )
15 return
16
17 type = mate1_type
18 joiner = fastqJoiner( type )
19 out_pairs = fastqWriter( open( outfile_pairs, 'wb' ), format = type )
20 out_singles = fastqWriter( open( outfile_singles, 'wb' ), format = type )
21
22 # Pairs + singles present in mate1
23 nof_singles = 0
24 nof_pairs = 0
25 mate2_input = fastqNamedReader( open( mate2_filename, 'rb' ), format = type )
26 i = None
27 for i, mate1 in enumerate( fastqReader( open( mate1_filename, 'rb' ), format = type ) ):
28 mate2 = mate2_input.get( joiner.get_paired_identifier( mate1 ) )
29 if mate2:
30 out_pairs.write( mate1 )
31 out_pairs.write( mate2 )
32 nof_pairs += 1
33 else:
34 out_singles.write( mate1 )
35 nof_singles += 1
36
37 # Singles present in mate2
38 mate1_input = fastqNamedReader( open( mate1_filename, 'rb' ), format = type )
39 j = None
40 for j, mate2 in enumerate( fastqReader( open( mate2_filename, 'rb' ), format = type ) ):
41 mate1 = mate1_input.get( joiner.get_paired_identifier( mate2 ) )
42 if not mate1:
43 out_singles.write( mate2 )
44 nof_singles += 1
45
46 if (i is None) and (j is None):
47 print "Your input files contained no valid FASTQ sequences."
48 else:
49 print 'There were %s single reads.' % ( nof_singles )
50 print 'Interlaced %s pairs of sequences.' % ( nof_pairs )
51
52 mate1_input.close()
53 mate2_input.close()
54 out_pairs.close()
55 out_singles.close()
56
57
58 if __name__ == "__main__":
59 main()