def __init__(
self,
collection,
analysis,
lsf=False,
max_guidetrees=10,
tmpdir=None,
datatype=None,
verbosity=0,
populate_cache=True,
debug=False,
):
optioncheck(analysis, ANALYSES + ['tc', 'TreeCollection'])
if analysis == 'tc':
self.analysis = 'TreeCollection'
else:
self.analysis = analysis
self.max_guidetrees = max_guidetrees
self.lsf = lsf
self.collection = collection
self.datatype = datatype or collection.datatype
self.verbosity = verbosity
optioncheck(self.datatype, ['protein', 'dna'])
self.tmpdir = tmpdir or collection.tmpdir
directorymake(self.tmpdir)
self.cache = {}
self.history = []
self.debug=debug
if populate_cache:
self.populate_cache()
def make_alf_dirs_(self):
""" DEPRECATED """
alf_dirs = {}
for k in range(self.num_classes):
dirname = fileIO.join_path(self.tmpdir, "class{0:0>1}".format(k + 1))
alf_dirs[k + 1] = errors.directorymake(dirname)
self.alf_dirs = alf_dirs
def __init__(
self,
class_list,
permutations_list,
nspecies,
subst_model,
rate_model,
master_tree_generator_method="yule",
master_tree=None,
class_tree_permuter="nni",
gene_length_kappa=1.7719,
gene_length_theta=279.9,
gene_length_min=10,
gamma_rate_param=None,
outdir="./",
autocorrelated_relaxed_clock=False,
uncorrelated_relaxed_clock=False,
scale_rates=False,
verbosity=0,
):
# default
errors.optioncheck(master_tree_generator_method, ["yule", "coal", "rtree", "custom"])
errors.optioncheck(class_tree_permuter, ["nni", "spr", "lgt", "genetree"])
if master_tree is None and master_tree_generator_method == "custom":
raise Exception("No custom tree was specified")
self.num_classes = len(class_list)
self.num_genes = sum(class_list)
self.class_list = class_list
self._master_tree = None
self.verbosity = verbosity
self.autocorrelated_relaxed_clock = autocorrelated_relaxed_clock
self.uncorrelated_relaxed_clock = uncorrelated_relaxed_clock
self.scale_rates = scale_rates
self.gene_trees = list()
if master_tree is None:
tree = self.generate_master_tree(master_tree_generator_method, nspecies)
self.master_tree = tree
self.num_species = nspecies
else:
self.master_tree = master_tree
if len(master_tree) != nspecies:
msg = [
"Warning: supplied tree has {0} taxa.".format(len(master_tree)),
"Required number is {0}.\n".format(nspecies),
"Resetting number of species to match the supplied tree.",
]
print("".join(msg))
self.num_species = nspecies
self.set_gene_lengths(gene_length_kappa, gene_length_theta, gene_length_min)
self.gamma_rate_param = gamma_rate_param
self.permuter = class_tree_permuter
self.permutations_list = permutations_list
self.datatype = datatype
self.tmpdir = errors.directorymake(tmpdir)
self.outdir = outdir
self.generate_class_trees() # sets self.class_trees dict
self.make_alf_dirs() # sets self.alf_dirs dict
self.write_alf_params()
self.get_true_partition()
def make_alf_dirs_(self):
""" DEPRECATED """
alf_dirs = {}
for k in range(self.num_classes):
dirname = fileIO.join_path(self.tmpdir,
'class{0:0>1}'.format(k + 1))
alf_dirs[k + 1] = errors.directorymake(dirname)
self.alf_dirs = alf_dirs
def write(self):
if hasattr(self, 'result'):
errors.directorymake(self.outdir)
errors.directorymake(fileIO.join_path(self.outdir,
'base_class_trees'))
errors.directorymake(fileIO.join_path(self.outdir,
'gene_trees'))
for rec in self.result:
filename = fileIO.join_path(self.outdir, rec.name) + '.phy'
rec.write_phylip(filename, interleaved=True)
for i in range(self.num_classes):
tree = self.class_trees[i+1]
name = 'base_tree_class{0:0>{1}}.nwk'.format(i+1,
len(str(self.num_classes)))
filename = fileIO.join_path(self.outdir, 'base_class_trees',
name)
tree.write_to_file(filename)
for i, tree in enumerate(self.gene_trees, start=1):
filename = fileIO.join_path(self.outdir, 'gene_trees',
tree.name+'.nwk')
tree.write_to_file(filename)
self.master_tree.write_to_file(fileIO.join_path(self.outdir,
'master_tree.nwk'))
filename = fileIO.join_path(self.outdir, 'true_partition.txt')
with open(filename, 'w') as partition_file:
partition_file.write(repr(self.true_partition))
def write(self):
if hasattr(self, 'result'):
errors.directorymake(self.outdir)
errors.directorymake(
fileIO.join_path(self.outdir, 'base_class_trees'))
errors.directorymake(fileIO.join_path(self.outdir, 'gene_trees'))
for rec in self.result:
filename = fileIO.join_path(self.outdir, rec.name) + '.phy'
rec.write_phylip(filename, interleaved=True)
for i in range(self.num_classes):
tree = self.class_trees[i + 1]
name = 'base_tree_class{0:0>{1}}.nwk'.format(
i + 1, len(str(self.num_classes)))
filename = fileIO.join_path(self.outdir, 'base_class_trees',
name)
tree.write_to_file(filename)
for i, tree in enumerate(self.gene_trees, start=1):
filename = fileIO.join_path(self.outdir, 'gene_trees',
tree.name + '.nwk')
tree.write_to_file(filename)
self.master_tree.write_to_file(
fileIO.join_path(self.outdir, 'master_tree.nwk'))
filename = fileIO.join_path(self.outdir, 'true_partition.txt')
with open(filename, 'w') as partition_file:
partition_file.write(repr(self.true_partition))
def __init__(
self,
records=None,
input_dir=None,
trees_dir=None,
file_format='fasta',
datatype=None,
tmpdir=TMPDIR,
calc_distances=False,
compression=None,
debug=False,
):
self.tmpdir = directorymake(tmpdir)
self._records = None
self.debug = debug
if records:
self.records = records
self.datatype = datatype or records[0].datatype
optioncheck(self.datatype, ['dna', 'protein'])
for rec in records:
rec.tmpdir = self.tmpdir
elif input_dir:
directorycheck(input_dir)
self.datatype = optioncheck(datatype, ['dna', 'protein'])
optioncheck(file_format, ['fasta', 'phylip'])
self.records = self.read_alignments(input_dir,
file_format,
compression)
else:
raise Exception('Provide a list of records, '
'or the path to a set of alignments')
self.taxon_set = TaxonSet()
if trees_dir:
self.read_trees(trees_dir, self.taxon_set)
if not self.records:
raise NoRecordsError(file_format, input_dir, compression)
if calc_distances:
self.calc_distances()
def write(self):
if hasattr(self, "result"):
errors.directorymake(self.outdir)
errors.directorymake(fileIO.join_path(self.outdir, "base_class_trees"))
errors.directorymake(fileIO.join_path(self.outdir, "gene_trees"))
for rec in self.result:
filename = fileIO.join_path(self.outdir, rec.name) + ".phy"
rec.write_phylip(filename, interleaved=True)
for i in range(self.num_classes):
tree = self.class_trees[i + 1]
name = "base_tree_class{0:0>{1}}.nwk".format(i + 1, len(str(self.num_classes)))
filename = fileIO.join_path(self.outdir, "base_class_trees", name)
tree.write_to_file(filename)
for i, tree in enumerate(self.gene_trees, start=1):
filename = fileIO.join_path(self.outdir, "gene_trees", tree.name + ".nwk")
tree.write_to_file(filename)
self.master_tree.write_to_file(fileIO.join_path(self.outdir, "master_tree.nwk"))
filename = fileIO.join_path(self.outdir, "true_partition.txt")
with open(filename, "w") as partition_file:
partition_file.write(repr(self.true_partition))
def __init__(
self,
class_list,
permutations_list,
nspecies,
tmpdir,
datatype='protein',
master_tree_generator_method='yule',
master_tree=None,
class_tree_permuter='nni',
gene_length_kappa=1.7719,
gene_length_theta=279.9,
gene_length_min=10,
gamma_rate_param=None,
outdir='./',
autocorrelated_relaxed_clock=False,
uncorrelated_relaxed_clock=False,
scale_rates=False,
verbosity=0,
):
# default
errors.optioncheck(master_tree_generator_method, ['yule', 'coal',
'rtree', 'custom'])
errors.optioncheck(class_tree_permuter, ['nni', 'spr', 'lgt', 'genetree'
])
if master_tree is None and master_tree_generator_method == 'custom':
raise Exception('No custom tree was specified')
self.num_classes = len(class_list)
self.num_genes = sum(class_list)
self.class_list = class_list
self.verbosity=verbosity
self.autocorrelated_relaxed_clock = autocorrelated_relaxed_clock
self.uncorrelated_relaxed_clock = uncorrelated_relaxed_clock
self.scale_rates = scale_rates
self.gene_trees = list()
if master_tree is None:
tree = self.generate_master_tree(master_tree_generator_method,
nspecies)
self.master_tree = tree
self.num_species = nspecies
else:
self.master_tree = master_tree
if len(master_tree) != nspecies:
msg = [
'Warning: supplied tree has {0} taxa.'.format(
len(master_tree)),
'Required number is {0}.\n'.format(nspecies),
'Resetting number of species to match the supplied tree.'
]
print(''.join(msg))
self.num_species = nspecies
self.set_gene_lengths(gene_length_kappa, gene_length_theta,
gene_length_min)
self.gamma_rate_param = gamma_rate_param
self.permuter = class_tree_permuter
self.permutations_list = permutations_list
self.datatype = datatype
self.tmpdir = errors.directorymake(tmpdir)
self.outdir = outdir
self.generate_class_trees() # sets self.class_trees dict
self.make_alf_dirs() # sets self.alf_dirs dict
self.write_alf_params()
self.get_true_partition()
def make_alf_dirs(self):
alf_dirs = {}
for i, g in enumerate(self.gene_trees, start=1):
dirname = fileIO.join_path(self.tmpdir, g.name)
alf_dirs[i] = errors.directorymake(dirname)
self.alf_dirs = alf_dirs
def __init__(
self,
class_list,
permutations_list,
nspecies,
subst_model,
rate_model,
master_tree_generator_method='yule',
master_tree=None,
class_tree_permuter='nni',
gene_length_kappa=1.7719,
gene_length_theta=279.9,
gene_length_min=10,
gamma_rate_param=None,
outdir='./',
autocorrelated_relaxed_clock=False,
uncorrelated_relaxed_clock=False,
scale_rates=False,
verbosity=0,
):
# default
errors.optioncheck(master_tree_generator_method,
['yule', 'coal', 'rtree', 'custom'])
errors.optioncheck(class_tree_permuter,
['nni', 'spr', 'lgt', 'genetree'])
if master_tree is None and master_tree_generator_method == 'custom':
raise Exception('No custom tree was specified')
self.num_classes = len(class_list)
self.num_genes = sum(class_list)
self.class_list = class_list
self._master_tree = None
self.verbosity = verbosity
self.autocorrelated_relaxed_clock = autocorrelated_relaxed_clock
self.uncorrelated_relaxed_clock = uncorrelated_relaxed_clock
self.scale_rates = scale_rates
self.gene_trees = list()
if master_tree is None:
tree = self.generate_master_tree(master_tree_generator_method,
nspecies)
self.master_tree = tree
self.num_species = nspecies
else:
self.master_tree = master_tree
if len(master_tree) != nspecies:
msg = [
'Warning: supplied tree has {0} taxa.'.format(
len(master_tree)),
'Required number is {0}.\n'.format(nspecies),
'Resetting number of species to match the supplied tree.'
]
print(''.join(msg))
self.num_species = nspecies
self.set_gene_lengths(gene_length_kappa, gene_length_theta,
gene_length_min)
self.gamma_rate_param = gamma_rate_param
self.permuter = class_tree_permuter
self.permutations_list = permutations_list
self.datatype = datatype
self.tmpdir = errors.directorymake(tmpdir)
self.outdir = outdir
self.generate_class_trees() # sets self.class_trees dict
self.make_alf_dirs() # sets self.alf_dirs dict
self.write_alf_params()
self.get_true_partition()
def make_alf_dirs(self):
alf_dirs = {}
for i, g in enumerate(self.gene_trees, start=1):
dirname = fileIO.join_path(self.tmpdir, g.name)
alf_dirs[i] = errors.directorymake(dirname)
self.alf_dirs = alf_dirs
