def test_insert_fully_resolved(self):
exp = self.final_multiple_fully_resolved
placements, backbone = parse_jplace(self.jplacedata_multiple)
obs = insert_fully_resolved(placements, backbone)
self.assertEqual({n.name
for n in obs.tips()}, {n.name
for n in exp.tips()})
self.assertEqual(obs.compare_rfd(exp), 0)
self.assertAlmostEqual(obs.compare_tip_distances(exp), 0)
def test_parse_jplace_multiple_per_fragment(self):
columns = [
'fragment', 'edge_num', 'likelihood', 'like_weight_ratio',
'distal_length', 'pendant_length'
]
exp_df = [
[
"82", 361, 0.01013206496780672, 1, 0.02652932626620403,
0.039354548684623215
],
[
"99", 308, 0.04520741687623886, 1, 0.11020044356641526,
0.06550337922097477
],
# tied on like_weight_ratio but lower pendant
[
"99", 309, 0.04520741687623886, 1, 0.11020044356641526,
0.00550337922097477
],
[
"55", 139, 0.09563944060686769, 1, 0.014593217782258146,
0.04537214236560885
],
# tied higher like_weight_ratio
[
"55", 138, 0.09563944060686769, 10, 0.014593217782258146,
0.04537214236560885
]
]
exp_df = pd.DataFrame(exp_df, columns=columns)
# ...adjust jplace data
data = json.loads(self.jplacedata)
keep = []
for placement in data['placements']:
if placement['n'][0] == '82':
keep.append(placement)
elif placement['n'][0] == '99':
placement['p'].append([
309, 0.04520741687623886, 1, 0.11020044356641526,
0.00550337922097477
])
keep.append(placement)
elif placement['n'][0] == '55':
placement['p'].append([
138, 0.09563944060686769, 10, 0.014593217782258146,
0.04537214236560885
])
keep.append(placement)
data['placements'] = keep
data = json.dumps(data)
exp_tree = self.tree
obs_df, obs_tree = parse_jplace(data)
obs_tree = to_skbio_treenode(obs_tree)
pdt.assert_frame_equal(obs_df, exp_df)
self.assertEqual(obs_tree.compare_rfd(exp_tree), 0)
def placement(placements, output, method):
if method == 'fully-resolved':
f = insert_fully_resolved
elif method == 'multifurcating':
f = insert_multifurcating
else:
raise ValueError("Unknown method: %s" % method)
placement_df, tree = parse_jplace(open(placements).read())
sktree = f(placement_df, tree)
sktree.write(output)
def test_place_jplace_square_braces(self):
self.jplacedata = json.loads(self.jplacedata)
treestr = self.jplacedata['tree']
treestr = re.sub(r"{(\d+)}", r"[\1]", treestr)
self.jplacedata['tree'] = treestr
self.jplacedata = json.dumps(self.jplacedata)
exp_tree = self.tree
obs_df, obs_tree = parse_jplace(self.jplacedata)
obs_tree = to_skbio_treenode(obs_tree)
self.assertEqual(obs_tree.compare_rfd(exp_tree), 0)
for n in obs_tree.traverse(include_self=False):
self.assertTrue(n.edge_num >= 0)
def test_parse_jplace_simple(self):
columns = [
'fragment', 'edge_num', 'likelihood', 'like_weight_ratio',
'distal_length', 'pendant_length'
]
exp_df = [[
"82", 361, 0.01013206496780672, 1, 0.02652932626620403,
0.039354548684623215
],
[
"99", 308, 0.04520741687623886, 1, 0.11020044356641526,
0.06550337922097477
],
[
"43", 309, 0.04054866161921744, 1, 0.010712923050783987,
0.020946988900520196
],
[
"195", 277, 0.01918907908397749, 1, 0.03065741838803451,
0.04513513498399864
],
[
"162", 55, 0.01758935282545493, 1, 0.0033199487685078776,
0.05388735804976052
],
[
"56", 81, 0.2366882303770561, 1, 0.04172580852519453,
0.0007060238727097983
],
[
"91", 105, 0.0001863393767883581, 1, 0.04578898721138839,
0.08655004339151215
],
[
"174", 89, 0.01216463967379211, 1, 0.04707020642820376,
0.045206727542450205
],
[
"5", 143, 0.012162345471765756, 1, 0.023797389484252734,
0.10447375403452556
],
[
"55", 139, 0.09563944060686769, 1, 0.014593217782258146,
0.04537214236560885
]]
exp_df = pd.DataFrame(exp_df, columns=columns)
exp_tree = self.tree
obs_df, obs_tree = parse_jplace(self.jplacedata)
obs_tree = to_skbio_treenode(obs_tree)
pdt.assert_frame_equal(obs_df, exp_df)
self.assertEqual(obs_tree.compare_rfd(exp_tree), 0)
def test_insert_fully_resolved_multiple_placements(self):
exp = self.final_multiple_fully_resolved
placements, backbone = parse_jplace(self.jplacedata_multiple)
# add another placement elsewhere that we would not keep
# as it's ratio is lower
dup1 = placements.iloc[0].copy()
dup1['like_weight_ratio'] -= 0.5
dup1['edge_num'] += 1
# add another placement elsewhere that we would not keep
# as, though its ratio is the same, its pendant is larger
dup2 = placements.iloc[1].copy()
dup2['pendant_length'] += 0.5
dup2['edge_num'] += 1
placements = pd.concat([placements, pd.DataFrame([dup1, dup2])])
obs = insert_fully_resolved(placements, backbone)
self.assertEqual({n.name
for n in obs.tips()}, {n.name
for n in exp.tips()})
self.assertEqual(obs.compare_rfd(exp), 0)
self.assertAlmostEqual(obs.compare_tip_distances(exp), 0)