/tools/fastq/fastq_masker_by_quality.py

 1#Dan Blankenberg
 2import string
 3from optparse import OptionParser
 4from galaxy_utils.sequence.fastq import fastqReader, fastqWriter
 5
 6
 7def get_score_comparer( operator ):
 8    if operator == 'gt':
 9        return compare_gt
10    elif operator == 'ge':
11        return compare_ge
12    elif operator == 'eq':
13        return compare_eq
14    elif operator == 'lt':
15        return compare_lt
16    elif operator == 'le':
17        return compare_le
18    elif operator == 'ne':
19        return compare_ne
20    raise 'Invalid operator provided: %s' % operator
21
22def compare_gt( quality_score, threshold_value ):
23    return quality_score > threshold_value
24
25def compare_ge( quality_score, threshold_value ):
26    return quality_score >= threshold_value
27
28def compare_eq( quality_score, threshold_value ):
29    return quality_score == threshold_value
30
31def compare_ne( quality_score, threshold_value ):
32    return quality_score != threshold_value
33
34def compare_lt( quality_score, threshold_value ):
35    return quality_score < threshold_value
36
37def compare_le( quality_score, threshold_value ):
38    return quality_score <= threshold_value
39
40class BaseReplacer( object ):
41    def __init__( self, replace_character ):
42        self.replace_character = replace_character
43    def __call__( self, base_character ):
44        return self.replace_character
45
46def main():
47    usage = "usage: %prog [options] input_file output_file"
48    parser = OptionParser( usage=usage )
49    parser.add_option( '-f', '--format', dest='format', type='choice', default='sanger', choices=( 'sanger', 'solexa', 'illumina' ), help='FASTQ variant type' )
50    parser.add_option( '-m', '--mask_character', dest='mask_character', default='N', help='Mask Character to use' )
51    parser.add_option( '-c', '--score_comparison', type="choice", dest='score_comparison', default='le', choices=('gt','ge','eq','lt', 'le', 'ne' ), help='Mask base when score is' )
52    parser.add_option( '-s', '--quality_score', type="float", dest='quality_score', default='0', help='Quality Score' )
53    parser.add_option( "-l", "--lowercase", action="store_true", dest="lowercase", default=False, help="Use lowercase masking")
54    ( options, args ) = parser.parse_args()
55    
56    if len ( args ) != 2:
57        parser.error( "Need to specify an input file and an output file" )
58    
59    score_comparer = get_score_comparer( options.score_comparison )
60    
61    if options.lowercase:
62        base_masker = string.lower
63    else:
64        base_masker = BaseReplacer( options.mask_character )
65    
66    out = fastqWriter( open( args[1], 'wb' ), format = options.format )
67    
68    num_reads = None
69    num_reads_excluded = 0
70    for num_reads, fastq_read in enumerate( fastqReader( open( args[0] ), format = options.format ) ):
71        sequence_list = list( fastq_read.sequence )
72        for i, quality_score in enumerate( fastq_read.get_decimal_quality_scores() ):
73            if score_comparer( quality_score, options.quality_score ):
74                sequence_list[ i ] = base_masker( sequence_list[ i ] )
75        fastq_read.sequence = "".join( sequence_list )
76        out.write( fastq_read )
77    
78    if num_reads is not None:
79        print "Processed %i %s reads." % ( num_reads + 1, options.format )
80    else:
81        print "No valid FASTQ reads were provided."
82
83if __name__ == "__main__": main()