def filter_phiX(self, reads):
filtered_reads = contaminants.pre_filter(self.phiX_bowtie2_index_prefix,
reads,
self.file_names['phiX_bam'],
)
for read in filtered_reads:
yield read
phiX_length_counts = sam.get_length_counts(self.file_names['phiX_bam'])
phiX_lengths = self.zero_padded_array(phiX_length_counts)
self.write_file('lengths', {'phiX': phiX_lengths})
def preprocess(self):
reads = self.get_reads()
trimmed_reads = self.trim_reads(reads)
rRNA_filtered_reads = contaminants.pre_filter(self.file_names['rRNA_index'],
trimmed_reads,
self.file_names['rRNA_bam'],
self.file_names['bowtie_error'],
)
with open(self.file_names['preprocessed_reads'], 'w') as preprocessed_fh:
for read in rRNA_filtered_reads:
preprocessed_fh.write(str(read))
