def test_from_kwargs():
"""Test NgdConfig initialises correctly from kwargs."""
config = NgdConfig.from_kwargs(parallel=2)
assert config.parallel == 2
with pytest.raises(ValueError):
NgdConfig.from_kwargs(garbage="wow")
def test_from_kwargs():
"""Test NgdConfig initialises correctly from kwargs."""
config = NgdConfig.from_kwargs(parallel=2)
assert config.parallel == 2
with pytest.raises(ValueError):
NgdConfig.from_kwargs(garbage="wow")
def test_get_choices():
"""Test NgdConfig.get_choices works."""
assert NgdConfig.get_choices('refseq_categories') == ['all', 'reference', 'representative', 'na']
with pytest.raises(KeyError):
NgdConfig.get_choices('garbage')
with pytest.raises(ValueError):
NgdConfig.get_choices('uri')
def test_get_choices():
"""Test NgdConfig.get_choices works."""
assert NgdConfig.get_choices('refseq_category') == ['all', 'reference', 'representative']
with pytest.raises(KeyError):
NgdConfig.get_choices('garbage')
with pytest.raises(ValueError):
NgdConfig.get_choices('uri')
def test_assembly_accessions():
"""Test NgdConfig.assembly_accessions getters/setters."""
config = NgdConfig()
assert config.assembly_accessions == []
config.assembly_accessions = "GCF_000203835.1"
assert config.assembly_accessions == ['GCF_000203835.1']
config.assembly_accessions = "GCF_000203835.1,GCF_000444875.1"
assert config.assembly_accessions == ['GCF_000203835.1', 'GCF_000444875.1']
def test_assembly_accessions():
"""Test NgdConfig.assembly_accessions getters/setters."""
config = NgdConfig()
assert config.assembly_accessions == []
config.assembly_accessions = "GCF_000203835.1"
assert config.assembly_accessions == ['GCF_000203835.1']
config.assembly_accessions = "GCF_000203835.1,GCF_000444875.1"
assert config.assembly_accessions == ['GCF_000203835.1', 'GCF_000444875.1']
def test_file_format():
"""Test NgdConfig.file_format getters/setters."""
config = NgdConfig()
assert config.file_format == ['genbank']
config.file_format = ['genbank', 'fasta']
assert config.file_format == ['genbank', 'fasta']
config.file_format = "all"
assert config.file_format == list(NgdConfig._FORMATS)
with pytest.raises(ValueError):
config.file_format = "garbage"
def test_group():
"""Test NgdConfig.group getters/setters."""
config = NgdConfig()
assert config.group == SUPPORTED_TAXONOMIC_GROUPS
config.group = ['bacteria', 'fungi']
assert config.group == ['bacteria', 'fungi']
config.group = "all"
assert config.group == SUPPORTED_TAXONOMIC_GROUPS
with pytest.raises(ValueError):
config.group = "garbage"
def test_file_formats():
"""Test NgdConfig.file_formats getters/setters."""
config = NgdConfig()
assert config.file_formats == ['genbank']
config.file_formats = ['genbank', 'fasta']
assert config.file_formats == ['genbank', 'fasta']
config.file_formats = "all"
assert config.file_formats == list(NgdConfig._FORMATS)
with pytest.raises(ValueError):
config.file_formats = "garbage"
def test_group():
"""Test NgdConfig.group getters/setters."""
config = NgdConfig()
assert config.group == SUPPORTED_TAXONOMIC_GROUPS
config.group = ['bacteria', 'fungi']
assert config.group == ['bacteria', 'fungi']
config.group = "all"
assert config.group == SUPPORTED_TAXONOMIC_GROUPS
with pytest.raises(ValueError):
config.group = "garbage"
def NCBI_get_info_GenbankID(data_folder,
acc_ID,
debug,
assembly_level_given='complete'):
section_given = get_section(acc_ID, debug)
if debug:
debug_message("-----------------------------------------")
debug_message("NCBI_get_info_GenbankID function call", color="yellow")
## import module and class
import ncbi_genome_download
from ncbi_genome_download.config import NgdConfig
tries = ['bacteria', 'archaea']
for entry_tried in tries:
if debug:
debug_message("Trying with: " + entry_tried, color="yellow")
ngd_config = NgdConfig.from_kwargs(section=section_given,
file_formats='genbank',
assembly_accessions=acc_ID,
output=data_folder,
dry_run=True,
groups=entry_tried)
info = ncbi_genome_download.core.select_candidates(ngd_config)
if info:
if debug:
debug_message("It worked!", color="yellow")
return (entry_tried)
raise "**** ERROR: Something happen while connecting to NCBI... ***"
exit()
return (False)
def NCBI_get_info(section_given,
data_folder,
tax_ID_list,
debug,
assembly_level_given='complete',
group_given='bacteria'):
'''This function uses ncbi_genome_download to
create a dry run and return information of each entry provided
'''
## import module and class
import ncbi_genome_download
from ncbi_genome_download.config import NgdConfig
try:
ngd_config = NgdConfig.from_kwargs(
section=section_given,
file_formats='genbank',
taxids=tax_ID_list,
output=data_folder,
dry_run=True,
assembly_levels=assembly_level_given,
groups=group_given)
info = ncbi_genome_download.core.select_candidates(ngd_config)
except:
raise "**** ERROR: Something happen while connecting to NCBI... ***"
exit()
return (False)
####
if (len(info)) < 1:
print(
colored(
"No entries matched your filter. Please check the input options provided",
'yellow'))
exit()
## fill dictionary to simplify
dict_entries = {}
for entry, _ in info:
strain_name = ncbi_genome_download.core.get_strain(entry)
## debug messagess
if debug:
debug_message("", 'yellow')
print(entry)
string = entry['assembly_accession'] + '\t' + entry[
'organism_name'] + '\t' + strain_name
debug_message(string, 'yellow')
debug_message(
".....................................................................\n",
'yellow')
## fill dictionary
dict_entries[entry['assembly_accession']] = (entry['organism_name'],
strain_name)
## return
return (dict_entries)
def test_is_compatible_assembly_level():
"""Test NgdConfig.is_compatible_assembly_level."""
config = NgdConfig()
ncbi_string = "Complete Genome"
assert config.is_compatible_assembly_level(ncbi_string)
config.assembly_levels = "complete"
assert config.is_compatible_assembly_level(ncbi_string)
config.assembly_levels = "chromosome,complete"
assert config.is_compatible_assembly_level(ncbi_string)
config.assembly_levels = "chromosome"
assert not config.is_compatible_assembly_level(ncbi_string)
def test_is_compatible_assembly_accession():
"""Test NgdConfig.is_compatible_assembly_accession."""
config = NgdConfig()
assert config.is_compatible_assembly_accession("GCF_000444875.1")
config.assembly_accessions = "GCF_000203835.1,GCF_000444875.1"
assert config.is_compatible_assembly_accession("GCF_000444875.1")
config.assembly_accessions = "GCF_000203835.1"
assert not config.is_compatible_assembly_accession("GCF_000444875.1")
def test_is_compatible_assembly_level():
"""Test NgdConfig.is_compatible_assembly_level."""
config = NgdConfig()
ncbi_string = "Complete Genome"
assert config.is_compatible_assembly_level(ncbi_string)
config.assembly_level = "complete"
assert config.is_compatible_assembly_level(ncbi_string)
config.assembly_level = "chromosome,complete"
assert config.is_compatible_assembly_level(ncbi_string)
config.assembly_level = "chromosome"
assert not config.is_compatible_assembly_level(ncbi_string)
def test_is_compatible_assembly_accession():
"""Test NgdConfig.is_compatible_assembly_accession."""
config = NgdConfig()
assert config.is_compatible_assembly_accession("GCF_000444875.1")
config.assembly_accessions = "GCF_000203835.1,GCF_000444875.1"
assert config.is_compatible_assembly_accession("GCF_000444875.1")
config.assembly_accessions = "GCF_000203835.1"
assert not config.is_compatible_assembly_accession("GCF_000444875.1")
def test_groups():
"""Test NgdConfig.groups getters/setters."""
config = NgdConfig()
assert config.groups == SUPPORTED_TAXONOMIC_GROUPS
config.groups = ['bacteria', 'fungi']
assert config.groups == ['bacteria', 'fungi']
config.groups = "all"
assert config.groups == SUPPORTED_TAXONOMIC_GROUPS
with pytest.raises(ValueError):
config.groups = "garbage"
# No metagenomes in refseq
with pytest.raises(ValueError):
config.section = "refseq"
config.groups = "metagenomes"
# genbank has metagenomes
config.section = "genbank"
config.groups = "metagenomes"
def test_init():
"""Test NgdConfig initialises with the correct default values."""
config = NgdConfig()
for key in NgdConfig._DEFAULTS:
expected = NgdConfig._DEFAULTS[key]
if key in NgdConfig._LIST_TYPES:
if expected:
if expected[0] == 'all':
expected = expected[1:]
else:
expected = expected[:1]
if 'all' in expected:
expected = expected[::].remove('all')
elif isinstance(expected, list):
expected = expected[0]
assert getattr(config, key) == expected
def test_assembly_levels():
"""Test NgdConfig.assembly_levels getters/setters."""
config = NgdConfig()
with pytest.raises(ValueError):
config.assembly_levels = 'garbage'
def test_section():
"""Test NgdConfig.section getters/setters."""
config = NgdConfig()
with pytest.raises(ValueError):
config.section = 'garbage'
def test_refseq_category():
"""Test NgdConfig.refseq_category getters/setters."""
config = NgdConfig()
with pytest.raises(ValueError):
config.refseq_category = 'garbage'
def test_from_namespace():
"""Test NgdConfig initialises correctly from a Namespace object."""
args = Namespace(parallel=2)
config = NgdConfig.from_namespace(args)
assert config.parallel == 2
def test_from_namespace():
"""Test NgdConfig initialises correctly from a Namespace object."""
args = Namespace(parallel=2)
config = NgdConfig.from_namespace(args)
assert config.parallel == 2
def test_type_materials():
"""Test NgdConfig.type_materials setters."""
config = NgdConfig()
with pytest.raises(ValueError):
config.type_materials = "invalid"
def test_refseq_categories():
"""Test NgdConfig.refseq_categories getters/setters."""
config = NgdConfig()
with pytest.raises(ValueError):
config.refseq_categories = 'garbage'
def test_section():
"""Test NgdConfig.section getters/setters."""
config = NgdConfig()
with pytest.raises(ValueError):
config.section = 'garbage'
def test_assembly_level():
"""Test NgdConfig.assembly_level getters/setters."""
config = NgdConfig()
with pytest.raises(ValueError):
config.assembly_level = 'garbage'