def get_task_parameters(args: argparse.Namespace, running_time) -> \
TaskParameters:
"""Returns TaskParameters object based on parsed arguments."""
return TaskParameters(
running_time=running_time,
multialignment_file_path=args.multialignment.filename,
multialignment_format=str(type(args.multialignment).__name__),
datatype=args.datatype.name,
metadata_file_path=args.metadata.filename if args.metadata else None,
blosum_file_path=args.blosum.filepath if args.blosum else None,
output_path=args.output_dir,
output_po=bool(args.output_po),
output_fasta=bool(args.output_fasta),
output_with_nodes=bool(args.output_full),
verbose=bool(args.verbose),
raw_maf=bool(args.raw_maf),
fasta_provider=args.fasta_provider
if args.fasta_provider else 'ConstSymbol',
cache=bool(args.cache),
missing_base_symbol=args.missing_symbol.value
if args.missing_symbol else missings.MissingBase().value,
fasta_source_file=str(args.fasta_path),
consensus_type=args.affinity,
hbmin=args.hbmin.value if args.hbmin else None,
stop=args.stop.value if args.stop else None,
p=args.p.value if args.p else None)
def setUp(self) -> None:
self.csv_files_dir = Path(__file__).parent.joinpath(
"csv_files").resolve()
self.alignment_files_dir = Path(__file__).parent.joinpath(
"alignment_files").resolve()
self.fasta_provider = missings.ConstBaseProvider(
missings.MissingBase())
def test_1_no_symbol_provided(self):
missing_symbol = missings.MissingBase()
const_symbol_provider = missings.ConstBaseProvider(missing_symbol)
expected_symbol = graph.Base('?')
actual_symbol = const_symbol_provider.get_base(msa.SequenceID('s'), 0)
self.assertEqual(expected_symbol, actual_symbol)
def test_3_incorrect_missing_symbol(self):
with self.assertRaises(ValueError) as e:
_ = missings.MissingBase('**')
expected_message = 'Missing symbol must be a single character.'
actual_message = str(e.exception)
self.assertEqual(expected_message, actual_message)
def build_from_dagmaf(maf: msa.Maf,
fasta_provider: Optional[missings.FastaProvider] = missings.ConstBaseProvider(
missings.MissingBase()),
metadata: Optional[msa.MetadataCSV] = None,
datatype: Optional[graph.DataType] = graph.DataType.Nucleotides) -> \
Tuple[graph.Poagraph, DAGMaf.DAGMaf]:
"""Converts MAF to DagMaf and builds poagraph from MAF file.
Args:
maf: Multialignment as MAF file.
fasta_provider: Provider of bases missing in DagMaf.
metadata: Metadata of sequences present in MAF.
datatype: Type of the processed data (nucleotides/proteins).
Returns:
Tuple: poagraph based on given input data and dagmaf created
from input MAF.
"""
dagmaf = maf2dagmaf.get_dagmaf(maf)
nodes, sequences = dagmaf2poagraph.get_poagraph(dagmaf,
fasta_provider,
metadata)
p = graph.Poagraph(nodes, sequences)
if metadata:
graph.Poagraph.complement_metadata_for_sequences_absent_in_metadata_provided(p, metadata)
p.datatype = datatype
return p, dagmaf
def setUp(self):
metadata_path = Path(__file__).parent.joinpath(
"../seq_metadata.csv").resolve()
self.metadatacsv = msa.MetadataCSV(
pathtools.get_file_content_stringio(metadata_path), metadata_path)
self.maf_files_dir = Path(__file__).parent.joinpath(
"maf_files_with_gaps").resolve()
self.missing_n = missings.MissingBase()
def resolve_fasta_provider(args: argparse.Namespace) -> \
missings.FastaProvider:
"""Returns fasta provider based on parsed arguments."""
if args.fasta_provider is None:
if args.missing_symbol is None:
return missings.ConstBaseProvider(missings.MissingBase())
else:
return missings.ConstBaseProvider(args.missing_symbol)
elif args.fasta_provider == 'ncbi':
use_cache = args.cache if args.cache else False
return missings.FromNCBI(use_cache)
elif args.fasta_provider == 'file':
if args.fasta_path is None:
raise Exception("""Fasta file source must be specified.
It must be provided when fasta source
is \'local\'.""")
return missings.FromFile(args.fasta_path)
else:
raise Exception("""Not known fasta provider.
Should be \'ncbi\' or \'file\' or None.
Cannot build graph.""")
def setUp(self):
metadata_path = Path(__file__).parent.joinpath("../seq_metadata.csv").resolve()
self.metadatacsv = msa.MetadataCSV(pathtools.get_file_content_stringio(metadata_path), metadata_path)
self.maf_files_dir = Path(__file__).parent.joinpath("maf_files_with_cycles_or_reversion").resolve()
self.fasta_provider = missings.ConstBaseProvider(missings.MissingBase())