/tools/fastq/fastq_paired_end_joiner.py

 1#Dan Blankenberg
 2import sys, os, shutil
 3from galaxy_utils.sequence.fastq import fastqReader, fastqNamedReader, fastqWriter, fastqJoiner
 4
 5def main():
 6    #Read command line arguments
 7    input1_filename = sys.argv[1]
 8    input1_type = sys.argv[2] or 'sanger'
 9    input2_filename = sys.argv[3]
10    input2_type = sys.argv[4] or 'sanger'
11    output_filename = sys.argv[5]
12    
13    if input1_type != input2_type:
14        print "WARNING: You are trying to join files of two different types: %s and %s." % ( input1_type, input2_type )
15    
16    input2 = fastqNamedReader( open( input2_filename, 'rb' ), input2_type )
17    joiner = fastqJoiner( input1_type )
18    out = fastqWriter( open( output_filename, 'wb' ), format = input1_type )
19    
20    i = None
21    skip_count = 0
22    for i, fastq_read in enumerate( fastqReader( open( input1_filename, 'rb' ), format = input1_type ) ):
23        identifier = joiner.get_paired_identifier( fastq_read )
24        fastq_paired = input2.get( identifier )
25        if fastq_paired is None:
26            skip_count += 1
27        else:
28            out.write( joiner.join( fastq_read, fastq_paired ) )
29    out.close()
30    
31    if i is None:
32        print "Your file contains no valid FASTQ reads."
33    else:
34        print input2.has_data()
35        print 'Joined %s of %s read pairs (%.2f%%).' % ( i - skip_count + 1, i + 1, float( i - skip_count + 1 ) / float( i + 1 ) * 100.0 )
36
37if __name__ == "__main__":
38    main()