def testTemplateIsOpenFile(self):
"""
BEAST2XML must run correctly when initialized from an open file
pointer.
"""
ET.fromstring(
BEAST2XML(template=StringIO(BEAST2XML().toString())).toString())
def testChainLength(self):
"""
Passing a chain length to toString must result in the expected XML.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
tree = ET.ElementTree(ET.fromstring(xml.toString(chainLength=100)))
elements = BEAST2XML.findElements(tree)
self.assertEqual('100', elements['run'].get('chainLength'))
def testDateUnit(self):
"""
Passing a dateUnit value to toString must result in the expected XML.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
tree = ET.ElementTree(ET.fromstring(xml.toString(dateUnit='day')))
elements = BEAST2XML.findElements(tree)
trait = elements['./run/state/tree/trait']
self.assertEqual('day', trait.get('units'))
def testScreenLogEvery(self):
"""
Passing a screenLogEvery value to toString must result in the expected
XML.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
tree = ET.ElementTree(ET.fromstring(xml.toString(screenLogEvery=300)))
elements = BEAST2XML.findElements(tree)
logger = elements["./run/logger[@id='screenlog']"]
self.assertEqual('300', logger.get('logEvery'))
def testDefaultDateDirection(self):
"""
Passing no dateDirection toString must result in the expected date
direction in the XML.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
tree = ET.ElementTree(ET.fromstring(xml.toString()))
elements = BEAST2XML.findElements(tree)
trait = elements['./run/state/tree/trait']
self.assertEqual('date-backward', trait.get('traitname'))
def testNoDateUnit(self):
"""
Passing no dateUnit value to toString must result in the expected XML
(with no 'units' attribute in the trait).
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
tree = ET.ElementTree(ET.fromstring(xml.toString()))
elements = BEAST2XML.findElements(tree)
trait = elements['./run/state/tree/trait']
self.assertEqual(None, trait.get('units'))
def testDefaultAge(self):
"""
Passing a default age to toString must result in the expected XML.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
xml.addSequence(Read('id1', 'ACTG'))
tree = ET.ElementTree(ET.fromstring(xml.toString(defaultAge=33.0)))
elements = BEAST2XML.findElements(tree)
# The sequence id with the default age of 0.0 must be in the traits.
trait = elements['./run/state/tree/trait']
self.assertTrue(trait.text.find('id1=33.0') > -1)
def testOneSequenceWithAgeAddedTogether(self):
"""
Adding a sequence with an age (both passed to addSequence) must result
in the expected XML.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
xml.addSequence(Read('id1', 'ACTG'), 44)
tree = ET.ElementTree(ET.fromstring(xml.toString()))
elements = BEAST2XML.findElements(tree)
# The sequence id with the given age must be in the traits.
trait = elements['./run/state/tree/trait']
self.assertTrue(trait.text.find('id1=44.0') > -1)
def testSequenceIdDateRegex(self):
"""
Using a sequence id date regex must result in the expected XML.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL,
sequenceIdDateRegex='^.*_([0-9]+)')
xml.addSequence(Read('id1_80_xxx', 'ACTG'))
tree = ET.ElementTree(ET.fromstring(xml.toString()))
elements = BEAST2XML.findElements(tree)
# The sequence id with the default age of 0.0 must be in the traits.
trait = elements['./run/state/tree/trait']
self.assertTrue(trait.text.find('id1_80_xxx=80.0') > -1)
def testSequenceIdDateRegexNonMatchingNotAnError(self):
"""
Using a sequence id date regex that doesn't match is not an error if
we pass sequenceIdDateRegexMayNotMatch=True, in which case the default
age should be assigned.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL,
sequenceIdDateRegex='^.*_([0-9]+)',
sequenceIdDateRegexMayNotMatch=True)
xml.addSequence(Read('id1_xxx', 'ACTG'))
tree = ET.ElementTree(ET.fromstring(xml.toString(defaultAge=50)))
elements = BEAST2XML.findElements(tree)
# The sequence id with the passed default age must be in the traits.
trait = elements['./run/state/tree/trait']
self.assertTrue(trait.text.find('id1_xxx=50.0') > -1)
def testTemplateWithNoRun(self, mock):
"""
Passing a template that has no <run> tag to BEAST2XML must raise a
ValueError when toString is called.
"""
xml = BEAST2XML(template='filename')
error = "^Could not find 'run' tag in XML template$"
assertRaisesRegex(self, ValueError, error, xml.toString)
def testLogFileBaseName(self):
"""
Passing a log file base name to toString must result in the expected
log file names in the XML.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
tree = ET.ElementTree(
ET.fromstring(xml.toString(logFileBasename='xxx')))
elements = BEAST2XML.findElements(tree)
logger = elements["./run/logger[@id='tracelog']"]
self.assertEqual('xxx' + BEAST2XML.TRACELOG_SUFFIX,
logger.get('fileName'))
logger = elements["./run/logger[@id='treelog.t:alignment']"]
self.assertEqual('xxx' + BEAST2XML.TREELOG_SUFFIX,
logger.get('fileName'))
def testAddSequences(self):
"""
Adding several sequences must result in the expected XML.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
xml.addSequences(
[Read('id1', 'GG'),
Read('id2', 'CC'),
Read('id3', 'AA')])
tree = ET.ElementTree(ET.fromstring(xml.toString()))
elements = BEAST2XML.findElements(tree)
# The sequences must be the children of the <data> tag.
data = elements['data']
self.assertEqual(3, len(data))
child = data[0]
self.assertEqual('sequence', child.tag)
self.assertEqual('GG', child.get('value'))
self.assertEqual('4', child.get('totalcount'))
self.assertEqual('id1', child.get('taxon'))
self.assertEqual('seq_id1', child.get('id'))
self.assertIs(None, child.text)
child = data[1]
self.assertEqual('sequence', child.tag)
self.assertEqual('CC', child.get('value'))
self.assertEqual('4', child.get('totalcount'))
self.assertEqual('id2', child.get('taxon'))
self.assertEqual('seq_id2', child.get('id'))
self.assertIs(None, child.text)
child = data[2]
self.assertEqual('sequence', child.tag)
self.assertEqual('AA', child.get('value'))
self.assertEqual('4', child.get('totalcount'))
self.assertEqual('id3', child.get('taxon'))
self.assertEqual('seq_id3', child.get('id'))
self.assertIs(None, child.text)
# The sequence ids with the default age of 0.0 must be in the traits.
trait = elements['./run/state/tree/trait']
self.assertTrue(trait.text.find('id1=0.0') > -1)
self.assertTrue(trait.text.find('id2=0.0') > -1)
self.assertTrue(trait.text.find('id3=0.0') > -1)
def testTemplateWithNoTracelog(self, mock):
"""
Passing a template that has no tracelog logger tag to BEAST2XML
must raise a ValueError when toString is called.
"""
xml = BEAST2XML(template='filename')
error = ('^Could not find "\./run/logger\[@id=\'tracelog\'\]" tag '
'in XML template$')
assertRaisesRegex(self, ValueError, error, xml.toString)
def testTemplateWithNoTrait(self, mock):
"""
Passing a template that has no <trait> tag to BEAST2XML must raise
a ValueError when toString is called.
"""
xml = BEAST2XML(template='filename')
error = ("^Could not find '\./run/state/tree/trait' tag in XML "
"template$")
assertRaisesRegex(self, ValueError, error, xml.toString)
def testDontMimicBEAUti(self):
"""
If mimicBEAUti is not passed to toString the BEAUti attributes must
not appear in the <beast> tag in the XML.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
tree = ET.ElementTree(ET.fromstring(xml.toString()))
root = tree.getroot()
self.assertEqual(None, root.get('beautitemplate'))
self.assertEqual(None, root.get('beautistatus'))
def testMimicBEAUti(self):
"""
Passing mimicBEAUti=True to toString must result in the expected
BEAUti attributes in the <beast> tag in the XML.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
tree = ET.ElementTree(ET.fromstring(xml.toString(mimicBEAUti=True)))
root = tree.getroot()
self.assertEqual('Standard', root.get('beautitemplate'))
self.assertEqual('', root.get('beautistatus'))
def testOneSequenceWithDateRegexAndDateUnitInDays(self):
"""
Adding a sequence with a date regex and date units in days must
result in the expected XML.
"""
sequenceDate = (date.today() - timedelta(days=10)).strftime('%Y-%m-%d')
r = r'^.*_(?P<year>\d\d\d\d)-(?P<month>\d\d)-(?P<day>\d\d)'
id_ = 'id1_' + sequenceDate
xml = BEAST2XML(clockModel=self.CLOCK_MODEL,
sequenceIdDateRegex=r,
dateUnit='day')
xml.addSequence(Read(id_, 'ACTG'))
tree = ET.ElementTree(ET.fromstring(xml.toString()))
elements = BEAST2XML.findElements(tree)
# The sequence id with an age of 10 days must be in the traits.
trait = elements['./run/state/tree/trait']
self.assertTrue(trait.text.find(id_ + '=10.0') > -1)
self.assertEqual('day', trait.get('units'))
def testExpectedTemplate(self):
"""
Passing a 'strict' clock model must result in the expected
XML template being loaded.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
tree = ET.ElementTree(ET.fromstring(xml.toString()))
root = tree.getroot()
logger = root.find(
'./run/logger[@id="tracelog"]/log[@idref="clockRate.c:alignment"]')
self.assertTrue(logger is not None)
def testSequenceIdDateRegexNonMatching(self):
"""
Using a sequence id date regex with a sequence id that does not match
must result in a ValueError.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL,
sequenceIdDateRegex='^.*_([0-9]+)')
error = ("^No sequence date could be found in 'id1' using the "
"sequence id date regex$")
assertRaisesRegex(self, ValueError, error, xml.addSequence,
Read('id1', 'ACTG'))
def testOneSequence(self):
"""
Adding a sequence must result in the expected XML.
"""
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
xml.addSequence(Read('id1', 'ACTG'))
tree = ET.ElementTree(ET.fromstring(xml.toString()))
elements = BEAST2XML.findElements(tree)
# The sequence must be the only child of the <data> tag.
data = elements['data']
self.assertEqual(1, len(data))
child = data[0]
self.assertEqual('sequence', child.tag)
self.assertEqual('ACTG', child.get('value'))
self.assertEqual('4', child.get('totalcount'))
self.assertEqual('id1', child.get('taxon'))
self.assertEqual('seq_id1', child.get('id'))
self.assertIs(None, child.text)
# The sequence id with the default age of 0.0 must be in the traits.
trait = elements['./run/state/tree/trait']
self.assertTrue(trait.text.find('id1=0.0') > -1)
def testTransformFunction(self):
"""
Passing a transform function to toString must result in the expected
XML.
"""
def transform(tree):
return ET.ElementTree(
ET.fromstring(
"<?xml version='1.0' encoding='UTF-8'?><hello/>"))
xml = BEAST2XML(clockModel=self.CLOCK_MODEL)
expected = ("<?xml version='1.0' encoding='" +
('UTF-8' if PY3 else 'utf-8') + "'?>\n<hello />")
self.assertEqual(expected, xml.toString(transformFunc=transform))
# backwards compatibility.
'--sequenceIdRegexMayNotMatch', '--sequenceIdDateRegexMayNotMatch',
action='store_false',
dest='sequenceIdRegexMustMatch',
help=('If specified (and --sequenceIdDateRegex or --sequenceIdAgeRegex is '
'given) it will not be considered an error if a sequence id does '
'not match the given regular expression. In that case, sequences '
'will be assigned an age of zero unless one is given via --age.'))
addFASTACommandLineOptions(parser)
args = parser.parse_args()
reads = parseFASTACommandLineOptions(args)
xml = BEAST2XML(
template=args.templateFile, clockModel=args.clockModel,
sequenceIdDateRegex=args.sequenceIdDateRegex,
sequenceIdAgeRegex=args.sequenceIdAgeRegex,
sequenceIdRegexMustMatch=args.sequenceIdRegexMustMatch,
dateUnit=args.dateUnit)
xml.addSequences(reads)
if args.age:
# Flatten lists of lists that we get from using both nargs='+' and
# action='append'. We use both because it allows people to use --age on the
# command line either via "--age id1=33 --age id2=21" or "--age id1=33
# id2=21", or a combination of these. That way it's not necessary to
# remember which way you're supposed to use it and you also can't be hit by
# the subtle problem encountered in
# https://github.com/acorg/dark-matter/issues/453
ages = list(chain.from_iterable(args.age))
def testNoArgsGivesValidXML(self):
"""
Passing no template or clock model to BEAST2XML and no arguments to
toString must produce valid XML.
"""
ET.fromstring(BEAST2XML().toString())
'match the id from its beginning.'))
parser.add_argument(
'--sequenceIdDateRegexMayNotMatch', action='store_true', default=False,
help=('If specified (and --sequenceIdDateRegex is given) it will not be '
'considered an error if a sequence id does not match the given '
'regular expression. In that case, sequences will be assigned the '
'default date unless one is given via --age.'))
addFASTACommandLineOptions(parser)
args = parser.parse_args()
reads = parseFASTACommandLineOptions(args)
xml = BEAST2XML(
template=args.templateFile, clockModel=args.clockModel,
sequenceIdDateRegex=args.sequenceIdDateRegex,
sequenceIdDateRegexMayNotMatch=args.sequenceIdDateRegexMayNotMatch,
)
xml.addSequences(reads)
if args.age:
# Flatten lists of lists that we get from using both nargs='+' and
# action='append'. We use both because it allows people to use --age on the
# command line either via "--age id1=33 --age id2=21" or "--age id1=33
# id2=21", or a combination of these. That way it's not necessary to
# remember which way you're supposed to use it and you also can't be hit by
# the subtle problem encountered in
# https://github.com/acorg/dark-matter/issues/453
ages = list(chain.from_iterable(args.age))
for ageInfo in ages: