[2] | 1 | #Dan Blankenberg |
---|
| 2 | import sys |
---|
| 3 | from galaxy_utils.sequence.fastq import fastqReader, fastqAggregator, fastqWriter |
---|
| 4 | |
---|
| 5 | def main(): |
---|
| 6 | input_filename = sys.argv[1] |
---|
| 7 | input_type = sys.argv[2] |
---|
| 8 | output_filename = sys.argv[3] |
---|
| 9 | output_type = sys.argv[4] |
---|
| 10 | force_quality_encoding = sys.argv[5] |
---|
| 11 | summarize_input = sys.argv[6] == 'summarize_input' |
---|
| 12 | if force_quality_encoding == 'None': |
---|
| 13 | force_quality_encoding = None |
---|
| 14 | |
---|
| 15 | aggregator = fastqAggregator() |
---|
| 16 | out = fastqWriter( open( output_filename, 'wb' ), format = output_type, force_quality_encoding = force_quality_encoding ) |
---|
| 17 | read_count = None |
---|
| 18 | for read_count, fastq_read in enumerate( fastqReader( open( input_filename ), format = input_type ) ): |
---|
| 19 | if summarize_input: |
---|
| 20 | aggregator.consume_read( fastq_read ) |
---|
| 21 | out.write( fastq_read ) |
---|
| 22 | out.close() |
---|
| 23 | |
---|
| 24 | if read_count is not None: |
---|
| 25 | print "Groomed %i %s reads into %s reads." % ( read_count + 1, input_type, output_type ) |
---|
| 26 | else: |
---|
| 27 | print "No valid FASTQ reads were provided." |
---|
| 28 | if input_type != output_type and 'solexa' in [ input_type, output_type ]: |
---|
| 29 | print "Converted between Solexa and PHRED scores." |
---|
| 30 | if summarize_input: |
---|
| 31 | print "Based upon quality and sequence, the input data is valid for: %s" % ( ", ".join( aggregator.get_valid_formats() ) or "None" ) |
---|
| 32 | ascii_range = aggregator.get_ascii_range() |
---|
| 33 | decimal_range = aggregator.get_decimal_range() |
---|
| 34 | print "Input ASCII range: %s(%i) - %s(%i)" % ( repr( ascii_range[0] ), ord( ascii_range[0] ), repr( ascii_range[1] ), ord( ascii_range[1] ) ) #print using repr, since \x00 (null) causes info truncation in galaxy when printed |
---|
| 35 | print "Input decimal range: %i - %i" % ( decimal_range[0], decimal_range[1] ) |
---|
| 36 | |
---|
| 37 | if __name__ == "__main__": main() |
---|