def test_bad_tree(self):
bad_trees = []
tree_file = "tests/data/species_trees/101g_nucl_conc_unconst.combined.nwk.tre"
with open(tree_file) as f:
bad_trees.append(f.read())
good_nexus = "#NEXUS\n\n"
good_nexus += "Begin taxa;\n"
good_nexus += " Dimensions ntax=3;\n"
good_nexus += " Taxlabels\n"
good_nexus += " spc01\n"
good_nexus += " spc02\n"
good_nexus += " spc03\n"
good_nexus += " ;\n"
good_nexus += "End;\n"
good_nexus += "Begin trees;\n"
good_nexus += " Translate\n"
good_nexus += " 1 spc01,\n"
good_nexus += " 2 spc02,\n"
good_nexus += " 3 spc03\n"
good_nexus += " ;\n"
good_nwk = "tree TREE1 = ((1[&dmv={0.1}]:1,2[&dmv={0.2}]:1)[&dmv={0.3}]"
good_nexus += "End;\n"
bad_trees.append(good_nexus.replace("#NEXUS", "#NEXU"))
bad_trees.append(good_nexus.replace("#NEXUS", "NEXUS"))
bad_trees.append(good_nexus.replace("tree TREE1", "tre TREE1"))
bad_trees.append(good_nexus.replace("End;", ""))
bad_trees.append(good_nexus.replace("Translate", "T"))
bad_trees.append(good_nexus.replace("2 spc02,", "2 spc02"))
bad_trees.append(good_nexus.replace("2 spc02,", "2 spc02 asdf,"))
bad_trees.append(good_nexus.replace("2 spc02,", "2 spc03,"))
bad_trees.append(good_nexus.replace("2 spc02,", "spc02 2,"))
bad_trees.append(good_nexus.replace("2 spc02,", "asdf2 spc02,"))
bad_trees.append(good_nexus.replace("2 spc02,", "spc02; 2,"))
bad_trees.append(good_nexus.replace(";\n", ""))
bad_trees.append(good_nexus.replace("Taxlabels", "Begin trees;"))
bad_trees.append(good_nexus.replace("dmv", "emv"))
bad_trees.append(good_nexus.replace("[", ""))
bad_trees.append(good_nexus.replace("[", "").replace("]", ""))
bad_trees.append(good_nexus.replace("=", ""))
bad_trees.append(good_nexus.replace("Begin taxa", "Begin trees"))
bad_trees.append(good_nexus.replace("Begin trees", "Begin taxa"))
bad_trees.append(good_nexus.replace("[&dmv={0.5}]", ""))
bad_trees.append(good_nexus.replace("[&dmv={0.1}]", ""))
bad_trees.append(good_nexus.replace("[&dmv={0.1}]", "[&dmv={asdf}]"))
bad_trees.append(good_nexus.replace(":1,2[&dmv", ":1, 2[&dmv"))
bad_trees.append(good_nexus.replace(good_nwk, good_nwk + good_nwk))
good_generation_time = 5
for bad_tree in bad_trees:
with pytest.raises(ValueError):
species_trees.parse_starbeast(
tree=bad_tree, generation_time=good_generation_time)
def verify(
self,
tree,
pop_size_map,
nexus=None,
branch_length_units="yr",
generation_time=1,
):
if nexus is None:
nexus = make_nexus(tree, pop_size_map)
demography = species_trees.parse_starbeast(nexus, generation_time,
branch_length_units)
assert demography.num_populations == tree.num_nodes
for pop in demography.populations:
assert pop.growth_rate == 0
# Population IDs are mapped to leaves first, and then to the internal nodes
# in postorder
pop_id_map = {}
k = 0
for u in tree.leaves():
pop_id_map[u] = k
k += 1
for u in tree.nodes(order="postorder"):
if tree.is_internal(u):
pop_id_map[u] = k
k += 1
for u in tree.nodes():
pop = demography.populations[pop_id_map[u]]
assert pop.growth_rate == 0
assert pop.initial_size == pop_size_map[u]
if tree.is_leaf(u):
# Note: we're assuming the default newick here in tskit that labels
# nodes as their id + 1.
assert pop.name == f"spc{u}"
else:
assert pop.name == f"pop_{pop_id_map[u]}"
# We should have demographic events for every internal node, and
# events should be output in increasing time order.
j = 0
for node in tree.nodes(order="timeasc"):
children = tree.children(node)
dest = pop_id_map[node]
for child in children:
event = demography.events[j]
j += 1
assert isinstance(event, msprime.MassMigration)
assert event.time == pytest.approx(tree.time(node))
source = pop_id_map[child]
assert event.source == source
assert event.dest == dest
assert j == len(demography.events)
def test_bad_parameter(self):
with open("tests/data/species_trees/91genes_species_rev.tre") as f:
good_tree = f.read()
good_time_units = "myr"
for bad_time_units in [-3, "asdf", ["myr"], "gen"]:
with pytest.raises(ValueError):
species_trees.parse_starbeast(
tree=f.read(),
time_units=bad_time_units,
generation_time=5,
)
for bad_generation_time in [-3, "sdf"]:
with pytest.raises(ValueError):
species_trees.parse_starbeast(
tree=good_tree,
time_units=good_time_units,
generation_time=bad_generation_time,
)
for bad_generation_time in [None, {}]:
with pytest.raises(TypeError):
species_trees.parse_starbeast(
tree=good_tree,
time_units=good_time_units,
generation_time=bad_generation_time,
)
def test_12_species(self):
with open("tests/data/species_trees/91genes_species_rev.tre") as f:
good_tree = f.read()
good_time_units = "myr"
good_generation_time = 5
spec = species_trees.parse_starbeast(
tree=good_tree,
time_units=good_time_units,
generation_time=good_generation_time,
)
assert len(spec.populations) == 23
for pop in spec.populations[:12]:
species_name = pop.name
assert species_name.startswith("spc")
assert species_name[3:].isnumeric()
assert len(spec.events) == 11
for mm in spec.events:
assert isinstance(mm, msprime.PopulationSplit)
def verify_non_ultrametric(self, tree, pop_size_map):
nexus = make_nexus(tree, pop_size_map)
with pytest.raises(ValueError):
species_trees.parse_starbeast(nexus, 10)