def _pcr_make_and_write_oligo_file_if_doesnt_exist(self):
if self.pcr_oligo_file_path is None:
oligo_file = [
f'forward\t{self.pcr_fwd_primer}',
f'reverse\t{self.pcr_rev_primer}'
]
self.pcr_oligo_file_path = os.path.join(self.input_dir, f'{self.name}_oligo_file.oligo')
write_list_to_destination(self.pcr_oligo_file_path, oligo_file)
def _make_contig_make_and_write_mothur_batch(self, dot_file_file_path):
mothur_batch_file = [
f'set.dir(input={self.input_dir})',
f'set.dir(output={self.output_dir})',
f'make.contigs(file={dot_file_file_path})'
]
self._set_mothur_batch_file_path()
write_list_to_destination(self.mothur_batch_file_path, mothur_batch_file)
def _dist_seqs_make_and_write_mothur_batch(self):
mothur_batch_file = [
f'set.dir(input={self.input_dir})',
f'set.dir(output={self.output_dir})',
f'dist.seqs(fasta={self.fasta_path}, countends=T, output=square)'
]
self._set_mothur_batch_file_path()
write_list_to_destination(self.mothur_batch_file_path, mothur_batch_file)
def _clearcut_make_and_write_mothur_batch(self):
mothur_batch_file = [
f'set.dir(input={self.input_dir})',
f'set.dir(output={self.output_dir})',
f'clearcut(phylip={self.dist_file_path}, verbose=t)'
]
self._set_mothur_batch_file_path()
write_list_to_destination(self.mothur_batch_file_path, mothur_batch_file)
def remove_primer_mismatch_annotations_from_fasta(fasta_path):
temp_fasta = []
fasta_to_clean = read_defined_file_to_list(fasta_path)
for i in range(len(fasta_to_clean) - 1):
if fasta_to_clean[i]:
if fasta_to_clean[i][0] == '>' and fasta_to_clean[i + 1]:
if '|' in fasta_to_clean[i]:
temp_fasta.extend([fasta_to_clean[i].split('|')[0], fasta_to_clean[i + 1]])
else:
temp_fasta.extend([fasta_to_clean[i].split('\t')[0], fasta_to_clean[i + 1]])
write_list_to_destination(fasta_path, temp_fasta)
def _weighted_unifrac_make_and_write_mothur_batch(self):
mothur_batch_file = [
f'set.dir(input={self.input_dir})',
f'set.dir(output={self.output_dir})',
f'unifrac.weighted(tree={self.tree_file_path}, group={self.group_file_path}, name={self.name_file_path},'
f' distance=square, processors={self.processors})'
]
self._set_mothur_batch_file_path()
write_list_to_destination(self.mothur_batch_file_path, mothur_batch_file)
def _split_abund_make_and_write_mothur_batch(self, abund_cutoff):
if self.name_file_path:
mothur_batch_file = [
f'set.dir(input={self.input_dir})',
f'set.dir(output={self.output_dir})',
f'split.abund(fasta={self.fasta_path}, name={self.name_file_path}, cutoff={abund_cutoff})'
]
else:
raise RuntimeError(
'Non name_file_path present. '
'A name file is necessary to be able to assess the abundances of sequences in the .fasta file'
)
self._set_mothur_batch_file_path()
write_list_to_destination(self.mothur_batch_file_path, mothur_batch_file)
def _summarise_make_and_write_mothur_batch(self):
if self.name_file_path:
mothur_batch_file = [
f'set.dir(input={self.input_dir})',
f'set.dir(output={self.output_dir})',
f'summary.seqs(fasta={self.fasta_path}, name={self.name_file_path})'
]
else:
mothur_batch_file = [
f'set.dir(input={self.input_dir})',
f'set.dir(output={self.output_dir})',
f'summary.seqs(fasta={self.fasta_path})'
]
self._set_mothur_batch_file_path()
write_list_to_destination(self.mothur_batch_file_path, mothur_batch_file)
def _if_scrap_fasta_exists_clean_and_write_out(self, fwd_output_scrapped_fasta_path):
# NB Mothur will not output a scrap fasta file if there are no scrap fasta. Also NB that mothur will output
# sequence names with no sequence for sequences that have multiple matches for a given primer.
# we should screen for these and remove them.
if fwd_output_scrapped_fasta_path == '':
return False
else:
scrapped_fasta_as_list = read_defined_file_to_list(fwd_output_scrapped_fasta_path)
if scrapped_fasta_as_list:
new_scrapped_fasta = self._make_new_fasta_no_multi_match_lines(scrapped_fasta_as_list)
if new_scrapped_fasta:
write_list_to_destination(fwd_output_scrapped_fasta_path, new_scrapped_fasta)
return True
else:
return False
else:
return False
def write_out_as_fasta(self):
self.file_path = self.infer_fasta_path_from_current_fastq_path()
write_list_to_destination(destination=self.file_path, list_to_write=self.as_fasta())
def _screen_seqs_make_and_write_mothur_batch_file(self, argument_dictionary):
mothur_batch_file = self._screen_seqs_make_mothur_batch_file(argument_dictionary)
self._set_mothur_batch_file_path()
write_list_to_destination(self.mothur_batch_file_path, mothur_batch_file)
def _pcr_make_and_write_mothur_batch_file(self):
mothur_batch_file = self._pcr_make_mothur_batch_file()
self._set_mothur_batch_file_path()
write_list_to_destination(self.mothur_batch_file_path, mothur_batch_file)
def _rev_comp_make_and_write_mothur_batch_file(self):
mothur_batch_file = self._make_rev_complement_mothur_batch_file()
self._set_mothur_batch_file_path()
write_list_to_destination(self.mothur_batch_file_path, mothur_batch_file)
def _make_contig_make_and_write_out_dot_file(self):
dot_file_file = [f'{self.fastq_gz_fwd_path} {self.fastq_gz_rev_path}']
dot_file_file_path = os.path.join(self.input_dir, f'{self.name}_fastq_pair.file')
write_list_to_destination(dot_file_file_path, dot_file_file)
return dot_file_file_path
