def testDiagnoseMulti(self):
multi_dir = data_source_path('testmulti/caenophidia')
fp = os.path.join(multi_dir, 'caenophidia_mos.fasta')
fp2 = os.path.join(multi_dir, 'caenophidia_mos2.fasta')
s = summary_stats_from_parse([fp, fp2], ["DNA", "RNA", "PROTEIN"],
careful_parse=False)
self.assertEqual(s[0], "PROTEIN")
self.assertEqual(s[1], [(114, 189), (109, 202)])
self.assertEqual(
s[2], 116
) # two taxa names were changed and 5 were deleted, so the union is 116
self.assertEqual(s[3], False)
fp3 = data_source_path('smallrna.fasta')
s = summary_stats_from_parse([fp3, fp3], ["DNA", "RNA", "PROTEIN"],
careful_parse=False)
self.assertEqual(s[0], "RNA")
self.assertEqual(s[1], [(32, 1650), (32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], True)
self.assertRaises(Exception,
summary_stats_from_parse, [fp, fp3],
["DNA", "RNA", "PROTEIN"],
careful_parse=False)
_LOG.warn(
"WARNING: summary_stats_from_parse will read multi with dna and protein as entirely protein. MIXED data type support is needed!"
)
fp4 = data_source_path('small.fasta')
fp5 = data_source_path('smallunaligned.fasta')
s = summary_stats_from_parse([fp4, fp4], ["DNA", "RNA", "PROTEIN"],
careful_parse=False)
self.assertEqual(s[0], "DNA")
self.assertEqual(s[1], [(32, 1650), (32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], True)
self.assertRaises(Exception,
summary_stats_from_parse, [fp, fp3],
["DNA", "RNA", "PROTEIN"],
careful_parse=False)
_LOG.warn(
"WARNING: summary_stats_from_parse will read multi with dna and protein as entirely protein. MIXED data type support is needed!"
)
fp4 = data_source_path('small.fasta')
fp5 = data_source_path('smallunaligned.fasta')
s = summary_stats_from_parse([fp4, fp5], ["DNA", "RNA", "PROTEIN"],
careful_parse=False)
self.assertEqual(s[0], "DNA")
self.assertEqual(s[1], [(32, 1650), (32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], False)
self.assertRaises(Exception,
summary_stats_from_parse, [fp, fp3],
["DNA", "RNA", "PROTEIN"],
careful_parse=False)
_LOG.warn(
"WARNING: summary_stats_from_parse will read multi with dna and protein as entirely protein. MIXED data type support is needed!"
)
def testDiagnoseMulti(self):
multi_dir = data_source_path("testmulti/caenophidia")
fp = os.path.join(multi_dir, "caenophidia_mos.fasta")
fp2 = os.path.join(multi_dir, "caenophidia_mos2.fasta")
s = summary_stats_from_parse([fp, fp2], ["DNA", "RNA", "PROTEIN"], careful_parse=False)
self.assertEqual(s[0], "PROTEIN")
self.assertEqual(s[1], [(114, 189), (109, 202)])
self.assertEqual(s[2], 116) # two taxa names were changed and 5 were deleted, so the union is 116
self.assertEqual(s[3], False)
fp3 = data_source_path("smallrna.fasta")
s = summary_stats_from_parse([fp3, fp3], ["DNA", "RNA", "PROTEIN"], careful_parse=False)
self.assertEqual(s[0], "RNA")
self.assertEqual(s[1], [(32, 1650), (32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], True)
self.assertRaises(
Exception, summary_stats_from_parse, [fp, fp3], ["DNA", "RNA", "PROTEIN"], careful_parse=False
)
_LOG.warn(
"WARNING: summary_stats_from_parse will read multi with dna and protein as entirely protein. MIXED data type support is needed!"
)
fp4 = data_source_path("small.fasta")
fp5 = data_source_path("smallunaligned.fasta")
s = summary_stats_from_parse([fp4, fp4], ["DNA", "RNA", "PROTEIN"], careful_parse=False)
self.assertEqual(s[0], "DNA")
self.assertEqual(s[1], [(32, 1650), (32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], True)
self.assertRaises(
Exception, summary_stats_from_parse, [fp, fp3], ["DNA", "RNA", "PROTEIN"], careful_parse=False
)
_LOG.warn(
"WARNING: summary_stats_from_parse will read multi with dna and protein as entirely protein. MIXED data type support is needed!"
)
fp4 = data_source_path("small.fasta")
fp5 = data_source_path("smallunaligned.fasta")
s = summary_stats_from_parse([fp4, fp5], ["DNA", "RNA", "PROTEIN"], careful_parse=False)
self.assertEqual(s[0], "DNA")
self.assertEqual(s[1], [(32, 1650), (32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], False)
self.assertRaises(
Exception, summary_stats_from_parse, [fp, fp3], ["DNA", "RNA", "PROTEIN"], careful_parse=False
)
_LOG.warn(
"WARNING: summary_stats_from_parse will read multi with dna and protein as entirely protein. MIXED data type support is needed!"
)
def testDiagnoseProt(self):
fp = data_source_path("caenophidia_mos.fasta")
print fp
s = summary_stats_from_parse([fp], ["DNA", "RNA", "PROTEIN"], careful_parse=False)
self.assertEqual(s[0], "PROTEIN")
self.assertEqual(s[1], [(114, 189)])
self.assertEqual(s[2], 114)
self.assertEqual(s[3], False)
s = summary_stats_from_parse([fp], ["DNA", "RNA", "PROTEIN"], careful_parse=True)
self.assertEqual(s[0], "PROTEIN")
self.assertEqual(s[1], [(114, 189)])
self.assertEqual(s[2], 114)
self.assertEqual(s[3], False)
self.assertRaises(Exception, summary_stats_from_parse, [fp], ["DNA", "RNA"], careful_parse=False)
self.assertRaises(Exception, summary_stats_from_parse, [fp], ["DNA", "RNA"], careful_parse=True)
def testDiagnoseRNA(self):
fp = data_source_path("smallrna.fasta")
print fp
s = summary_stats_from_parse([fp], ["DNA", "RNA", "PROTEIN"], careful_parse=False)
self.assertEqual(s[0], "RNA")
self.assertEqual(s[1], [(32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], True)
s = summary_stats_from_parse([fp], ["DNA", "RNA", "PROTEIN"], careful_parse=True)
self.assertEqual(s[0], "RNA")
self.assertEqual(s[1], [(32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], True)
self.assertRaises(Exception, summary_stats_from_parse, [fp], ["DNA", "PROTEIN"], careful_parse=False)
_LOG.warn("WARNING: summary_stats_from_parse does not distinguish between RNA and DNA in 'careful' mode")
def testDiagnoseDNA(self):
fp = data_source_path("small.fasta")
print fp
s = summary_stats_from_parse([fp], ["DNA", "RNA", "PROTEIN"], careful_parse=False)
self.assertEqual(s[0], "DNA")
self.assertEqual(s[1], [(32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], True)
s = summary_stats_from_parse([fp], ["DNA", "RNA", "PROTEIN"], careful_parse=True)
self.assertEqual(s[0], "DNA")
self.assertEqual(s[1], [(32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], True)
self.assertRaises(Exception, summary_stats_from_parse, [fp], ["RNA"], careful_parse=False)
self.assertRaises(Exception, summary_stats_from_parse, [fp], ["RNA"], careful_parse=True)
def populate_auto_options(user_config, md, force=False):
if user_config.commandline.input is None:
sys.exit("ERROR: Input file(s) not specified.")
from pasta.usersettingclasses import get_list_of_seq_filepaths_from_dir
from pasta.alignment import summary_stats_from_parse
try:
if user_config.commandline.multilocus:
fn_list = get_list_of_seq_filepaths_from_dir(
user_config.commandline.input)
else:
fn_list = [user_config.commandline.input]
datatype_list = [user_config.commandline.datatype.upper()]
careful_parse = user_config.commandline.untrusted
summary_stats = summary_stats_from_parse(fn_list,
datatype_list,
md,
careful_parse=careful_parse)
except:
if user_config.commandline.auto:
MESSENGER.send_error(
"Error reading input while setting options for the --auto mode\n"
)
else:
MESSENGER.send_error("Error reading input\n")
raise
if force or user_config.commandline.auto:
user_config.commandline.auto = False
auto_opts = get_auto_defaults_from_summary_stats(
summary_stats[0], summary_stats[1], summary_stats[2])
user_config.get('sate').set_values_from_dict(auto_opts['sate'])
user_config.get('commandline').set_values_from_dict(
auto_opts['commandline'])
user_config.get('fasttree').set_values_from_dict(auto_opts['fasttree'])
def populate_auto_options(user_config, md, force=False):
if user_config.commandline.input is None:
sys.exit("ERROR: Input file(s) not specified.")
from pasta.usersettingclasses import get_list_of_seq_filepaths_from_dir
from pasta.alignment import summary_stats_from_parse
try:
if user_config.commandline.multilocus:
fn_list = get_list_of_seq_filepaths_from_dir(user_config.commandline.input)
else:
fn_list = [user_config.commandline.input]
datatype_list = [user_config.commandline.datatype.upper()]
careful_parse = user_config.commandline.untrusted
summary_stats = summary_stats_from_parse(fn_list, datatype_list, md, careful_parse=careful_parse)
except:
if user_config.commandline.auto:
MESSENGER.send_error("Error reading input while setting options for the --auto mode\n")
else:
MESSENGER.send_error("Error reading input\n")
raise
if force or user_config.commandline.auto:
user_config.commandline.auto = False
auto_opts = get_auto_defaults_from_summary_stats(summary_stats[0], summary_stats[1], summary_stats[2])
user_config.get('sate').set_values_from_dict(auto_opts['sate'])
user_config.get('commandline').set_values_from_dict(auto_opts['commandline'])
user_config.get('fasttree').set_values_from_dict(auto_opts['fasttree'])
def testDiagnoseProt(self):
fp = data_source_path('caenophidia_mos.fasta')
print(fp)
s = summary_stats_from_parse([fp], ["DNA", "RNA", "PROTEIN"],
careful_parse=False)
self.assertEqual(s[0], "PROTEIN")
self.assertEqual(s[1], [(114, 189)])
self.assertEqual(s[2], 114)
self.assertEqual(s[3], False)
s = summary_stats_from_parse([fp], ["DNA", "RNA", "PROTEIN"],
careful_parse=True)
self.assertEqual(s[0], "PROTEIN")
self.assertEqual(s[1], [(114, 189)])
self.assertEqual(s[2], 114)
self.assertEqual(s[3], False)
self.assertRaises(Exception,
summary_stats_from_parse, [fp], ["DNA", "RNA"],
careful_parse=False)
self.assertRaises(Exception,
summary_stats_from_parse, [fp], ["DNA", "RNA"],
careful_parse=True)
def testDiagnoseRNA(self):
fp = data_source_path('smallrna.fasta')
print(fp)
s = summary_stats_from_parse([fp], ["DNA", "RNA", "PROTEIN"],
careful_parse=False)
self.assertEqual(s[0], "RNA")
self.assertEqual(s[1], [(32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], True)
s = summary_stats_from_parse([fp], ["DNA", "RNA", "PROTEIN"],
careful_parse=True)
self.assertEqual(s[0], "RNA")
self.assertEqual(s[1], [(32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], True)
self.assertRaises(Exception,
summary_stats_from_parse, [fp], ["DNA", "PROTEIN"],
careful_parse=False)
_LOG.warn(
"WARNING: summary_stats_from_parse does not distinguish between RNA and DNA in 'careful' mode"
)
def testDiagnoseDNA(self):
fp = data_source_path('small.fasta')
print(fp)
s = summary_stats_from_parse([fp], ["DNA", "RNA", "PROTEIN"],
careful_parse=False)
self.assertEqual(s[0], "DNA")
self.assertEqual(s[1], [(32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], True)
s = summary_stats_from_parse([fp], ["DNA", "RNA", "PROTEIN"],
careful_parse=True)
self.assertEqual(s[0], "DNA")
self.assertEqual(s[1], [(32, 1650)])
self.assertEqual(s[2], 32)
self.assertEqual(s[3], True)
self.assertRaises(Exception,
summary_stats_from_parse, [fp], ["RNA"],
careful_parse=False)
self.assertRaises(Exception,
summary_stats_from_parse, [fp], ["RNA"],
careful_parse=True)
