def contig_collection_a5_repeat() -> cnt.ContigCollection:
mink: int = 15
maxk: int = 15
contig_collection: cnt.ContigCollection = cnt.get_contig_collection(
os.path.join('tests', 'data', 'test_contigs_a5_repeat.fasta'), maxk)
overlap_collection: ovl.OverlapCollection = ovl.detect_adjacent_contigs(
contig_collection, mink, maxk)
return contig_collection, overlap_collection
def contig_collection_spades_1() -> cnt.ContigCollection:
mink: int = 8
maxk: int = 17
contig_collection: cnt.ContigCollection = cnt.get_contig_collection(
os.path.join('tests', 'data', 'test_contigs_spades_1.fasta.gz'), maxk)
overlap_collection: ovl.OverlapCollection = ovl.detect_adjacent_contigs(
contig_collection, mink, maxk)
return contig_collection, overlap_collection
def contig_collection_spades_no_multplty_0() -> cnt.ContigCollection:
mink: int = 16
maxk: int = 25
contig_collection: cnt.ContigCollection = cnt.get_contig_collection(
os.path.join('tests', 'data',
'test_contigs_spades_no_multplty_0.fasta'), maxk)
overlap_collection: ovl.OverlapCollection = ovl.detect_adjacent_contigs(
contig_collection, mink, maxk)
return contig_collection, overlap_collection
def _do_combinator_work(infpath: str, mink: int,
maxk: int) -> MockContigsFixture:
# Read contigs
contig_collection: cnt.ContigCollection = cnt.get_contig_collection(
infpath, maxk)
# Detect adjacent contigs
overlap_collection: ovl.OverlapCollection = ovl.detect_adjacent_contigs(
contig_collection, mink, maxk)
# Assign multiplicity to contigs
amu.assign_multiplty(contig_collection, overlap_collection)
return contig_collection, overlap_collection
def main(version: str, last_update_date: str) -> None:
contigs_fpaths: Sequence[str] # paths to input files
params: Dict[str, Any] # parameters of the program
# Parse command line arguments
contigs_fpaths, params = parse_args(version, last_update_date)
# Report parameters of current run
_report_parameters(params, version, last_update_date)
# Iterate over input files and process them
fpath: str
for fpath in contigs_fpaths:
print('Processing file `{}`'.format(fpath))
# Create output dir
make_outdir(params['o'])
# Read contigs
contig_collection: cnt.ContigCollection = cnt.get_contig_collection(
fpath, params['a'])
# Detect adjacent contigs
overlap_collection: ovl.OverlapCollection = ovl.detect_adjacent_contigs(
contig_collection, params['i'], params['a'])
# Assign multiplicity to contigs
amu.assign_multiplty(contig_collection, overlap_collection)
# Make prefix for current input file
prefix: str = conf_prefix(fpath, params['o'])
# Write output files
# Write adjacency table
out.write_adjacency_table(contig_collection, overlap_collection,
params['o'], prefix)
# Write full log
out.write_full_log(contig_collection, overlap_collection, params['o'],
prefix)
# Write summary
out.write_summary(contig_collection, overlap_collection, fpath,
params['o'], prefix)
print('-' * 20)