def _align_paired_end_reads(self):
"""Manage the actual alignemnt of paired end reads."""
read_aligner = Segemehl(self._args.segemehl_bin, self._args.progress)
if self._helpers.file_needs_to_be_created(self._paths.index_path):
read_aligner.build_index(self._paths.ref_seq_paths,
self._paths.index_path)
for read_path_pair, output_path, nomatch_path, bam_path in zip(
self._paths.processed_read_path_pairs,
self._paths.primary_read_aligner_sam_paths,
self._paths.unaligned_reads_paths,
self._paths.primary_read_aligner_bam_paths):
if not self._helpers.file_needs_to_be_created(output_path):
continue
elif not self._helpers.file_needs_to_be_created(bam_path):
continue
read_aligner.run_alignment(read_path_pair,
self._paths.index_path,
self._paths.ref_seq_paths,
output_path,
int(self._args.processes),
nomatch_path,
int(self._args.hit_strategy),
int(self._args.segemehl_accuracy),
float(self._args.segemehl_evalue),
self._args.split,
paired_end=True)
class ReadAligner(object):
"""An abstraction layer for different short read aligners."""
def __init__(self, segemehl_bin, show_progress):
self.segemehl = Segemehl(segemehl_bin, show_progress=show_progress)
def build_index(self, ref_seq_paths, index_path):
self.segemehl.build_index(ref_seq_paths, index_path)
def run_alignment(
self,
read_path_or_pair,
index_path,
ref_seq_path,
output_path,
nomatch_path,
threads,
accuracy,
evalue,
split,
paired_end=False,
):
self.segemehl.align_reads(
read_path_or_pair,
index_path,
ref_seq_path,
output_path,
nonmatch_file=nomatch_path,
threads=threads,
accuracy=accuracy,
evalue=evalue,
split=split,
paired_end=paired_end,
)
class ReadAligner(object):
"""An abstraction layer for different short read aligners."""
def __init__(self, segemehl_bin, show_progress):
self.segemehl = Segemehl(segemehl_bin, show_progress=show_progress)
def build_index(self, ref_seq_paths, index_path):
self.segemehl.build_index(ref_seq_paths, index_path)
def run_alignment(self, read_path_or_pair, index_path, ref_seq_path, output_path,
nomatch_path, threads, accuracy, evalue, split, paired_end=False):
self.segemehl.align_reads(read_path_or_pair, index_path, ref_seq_path,
output_path, nonmatch_file=nomatch_path,
threads=threads, accuracy=accuracy,
evalue=evalue, split=split, paired_end=paired_end)
def _align_single_end_reads(self):
"""Manage the actual alignment of single end reads."""
read_aligner = Segemehl(
self._args.segemehl_bin, self._args.progress)
if self._file_needs_to_be_created(self._paths.index_path):
read_aligner.build_index(
self._paths.ref_seq_paths, self._paths.index_path)
for read_path, output_path, nomatch_path, bam_path in zip(
self._paths.processed_read_paths,
self._paths.primary_read_aligner_sam_paths,
self._paths.unaligned_reads_paths,
self._paths.read_alignment_bam_paths):
if not self._file_needs_to_be_created(output_path):
continue
elif not self._file_needs_to_be_created(bam_path):
continue
read_aligner.run_alignment(
read_path, self._paths.index_path, self._paths.ref_seq_paths,
output_path, int(self._args.processes), nomatch_path,
int(self._args.hit_strategy),
int(self._args.segemehl_accuracy),
float(self._args.segemehl_evalue), self._args.split,
paired_end=False)
