def test_replacing_hgt(self):
N = 20
S = te.simulate_species_tree(N, model='innovation')
# true gene tree (with losses)
TGT = te.simulate_dated_gene_tree(
S,
dupl_rate=0.0,
loss_rate=0.0,
hgt_rate=1.0,
prohibit_extinction='per_species',
replace_prob=1.0,
)
# observable gene tree
OGT = te.observable_tree(TGT)
leaves = [v for v in OGT.leaves()]
colors = {v.color for v in leaves}
# print(TGT.to_newick())
# print(OGT.to_newick())
self.assertTrue(len(colors) == N and len(leaves) == N)
def test_rs_edges(self):
S = te.simulate_species_tree(10)
TGT = te.simulate_dated_gene_tree(S,
dupl_rate=1.0,
loss_rate=0.5,
hgt_rate=0.5)
OGT = te.observable_tree(TGT)
transf1 = analysis.true_transfer_edges(OGT)
transf2 = analysis.rs_transfer_edges(OGT, S)
self.assertTrue(transf1.issuperset(transf2))
def generate_solutions_unique_species(n, i_p=0.5, d_p=0.5):
done = False
count = 0
while not done:
S = te.simulate_species_tree(10, model='innovation')
TGT = te.simulate_dated_gene_tree(S,
dupl_rate=0.5,
loss_rate=0.5,
hgt_rate=0.5,
prohibit_extinction="per_family",
replace_prob=0.0)
OGT = te.observable_tree(TGT)
ldt = ldt_graph(OGT, S)
if len(ldt.nodes()) == n:
IG = InvestigateGraph(ldt)
IG.perturb_graph(i_p, d_p)
solver = LDTEditor(IG._G_perturbed)
solver.build_model()
solver.optimize(time_limit=None)
sol_graph, sol_distance = solver.get_solution()
properly_colored = is_properly_colored(sol_graph)
cograph = is_cograph(sol_graph)
compatible = is_compatible(sol_graph)
edit_dist = gt.symmetric_diff(IG._G_perturbed, sol_graph)
print("Runtime: {}".format(solver.get_solve_time()))
if properly_colored and cograph and compatible:
print("Saving data...")
solver._save_ILP_data(
IG._G_perturbed,
sol_graph,
solver.get_solve_time(),
edit_dist,
only_add=False,
only_delete=False,
filename="{}nodes/LDTEdit_exact_solution".format(n))
else:
print("No solution found!")
count += 1
if count == 100:
done = True
def generate_trees(n=100,
m=10,
model='innovation',
dupl_rate=0.5,
loss_rate=0.5,
hgt_rate=0.5,
prohibit_extinction="per_family",
replace_prob=0.0,
size=10):
i = 0
dirName = 'exact_solutions/trees/{}trees'.format(size)
# create folder if it doesnt exist
if not os.path.exists(dirName):
os.makedirs(dirName)
ID = 0
else:
ID = find_next_ID('exact_solutions/trees/{}trees/'.format(size))
while i < n:
S = te.simulate_species_tree(m, model=model)
TGT = te.simulate_dated_gene_tree(
S,
dupl_rate=dupl_rate,
loss_rate=loss_rate,
hgt_rate=hgt_rate,
prohibit_extinction=prohibit_extinction,
replace_prob=replace_prob)
OGT = te.observable_tree(TGT)
ldt = ldt_graph(OGT, S)
amount_nodes = len(ldt.nodes())
if amount_nodes == size:
# save trees
filename_species = 'exact_solutions/trees/{}trees/species_{}_{}_{}.json'.format(
size, m, model, ID)
filename_gene = 'exact_solutions/trees/{}trees/gene_{}_{}_{}_{}_{}_{}.json'.format(
size, dupl_rate, loss_rate, hgt_rate, prohibit_extinction,
replace_prob, ID)
S.serialize(filename_species)
TGT.serialize(filename_gene)
ID += 1
i += 1
def test_ldt_fitch(self):
S = te.simulate_species_tree(20, model='innovation')
# true gene tree (with losses)
TGT = te.simulate_dated_gene_tree(S,
dupl_rate=1.0,
loss_rate=0.5,
hgt_rate=0.2)
# observable gene tree
OGT = te.observable_tree(TGT)
# finally we can extract the LDT and Fitch graph
ldt = analysis.ldt_graph(OGT, S)
transfer_edges = analysis.rs_transfer_edges(OGT, S)
fitch = analysis.undirected_fitch(OGT, transfer_edges)
cotree = to_cotree(ldt)
self.assertTrue(gt.is_subgraph(ldt, fitch) and cotree)
def simulate_gene_trees(self, N, **kwargs):
self.number_of_families = N
self.true_gene_trees = te.simulate_gene_trees(self.S, N=N, **kwargs)
if N == 1:
self.true_gene_trees = [self.true_gene_trees]
self.observable_gene_trees = [
te.observable_tree(tree) for tree in self.true_gene_trees
]
# sequences should be emptied here if methods were called before
if hasattr(self, 'sequence_dicts'):
self.sequence_dicts.clear()
if self.outdir:
for i in range(N):
filename = self._path('true_gene_trees',
'gene_tree{}.json'.format(i))
self.true_gene_trees[i].serialize(filename, mode='json')
# -*- coding: utf-8 -*-
import tralda.tools.GraphTools as gt
import asymmetree.treeevolve as te
from asymmetree.analysis import (undirected_fitch,
rs_transfer_edges,
below_equal_above,
ldt_graph,
RsScenarioConstructor,)
from asymmetree.tools.PhyloTreeTools import (to_newick,)
S = te.simulate_species_tree(10)
TGT = te.simulate_dated_gene_tree(S, dupl_rate=1.0, loss_rate=0.5,
hgt_rate=0.5)
OGT = te.observable_tree(TGT)
print('--- S ---\n', to_newick(S))
print(to_newick(S, distance=False, label_inner=False))
print('--- OGT ---\n', to_newick(OGT))
ldt, above, equal = below_equal_above(OGT, S)
fitch = undirected_fitch(OGT, rs_transfer_edges(OGT, S))
n = ldt.order()
print('Genes:', n, 'Total relations:', int(n * (n-1) / 2))
print('< {}\n= {}\n> {}'.format(ldt.size(), equal.size(), above.size()))
rs_scen_constr = RsScenarioConstructor(ldt)
result = rs_scen_constr.run()
if result:
# build loop
len(parameter_Df.index)
for ind in range(len(parameter_Df.index)-1):
# species tree of type ’PhyloTree’
s = te.simulate_species_tree(int(parameter_Df.loc[ind, 'num_of_leaves']),
model = parameter_Df.loc[ind, 'model'],
non_binary_prob = parameter_Df.loc[ind, 'non_binary_prob'],
planted = parameter_Df.loc[ind, 'planted'],
remove_extinct = parameter_Df.loc[ind, 'remove_extinct'],
rescale_to_height = parameter_Df.loc[ind, 'rescale_to_height']
)
# true gene tree (contains losses) of type ’PhyloTree’
tgt = te.simulate_dated_gene_tree(s,
dupl_rate = parameter_Df.loc[ind, 'dupl_rate'],
loss_rate = parameter_Df.loc[ind, 'loss_rate'],
hgt_rate = parameter_Df.loc[ind, 'hgt_rate'],
dupl_polytomy = 0.0,
prohibit_extinction= parameter_Df.loc[ind, 'prohibit_extinction'],
replace_prob = parameter_Df.loc[ind, 'replace_prob']
)
# serialization
s.serialize(wk_dir / '01_Data' / str(parameter_Df.loc[ind, 'ID'] + '_species_tree.pickle'))
tgt.serialize(wk_dir / '01_Data' / str(parameter_Df.loc[ind, 'ID'] + '_gene_tree.pickle'))
print('Simulating Tree :', ind)
ogt = te.observable_tree(tgt)
def benchmark_fromTrees(n, p1, p2, filename='exact_solutions/trees'):
# load species+gene trees
name = filename + '/{}trees'.format(n)
tree_files = []
#probabilities = [0.15, 0.30, 0.50]
#probs = [(0.15, 0.15), (0.3, 0.3), (0.5, 0.5), (0.15, 0.5), (0.5, 0.15)]
#nodes = [10, 14, 18]
for _, _, files in os.walk(name):
for file in files:
tree_files.append(file)
species_trees = []
gene_trees = []
for f in tree_files:
if f.startswith('species'):
species_trees.append(f)
else:
gene_trees.append(f)
c1_graphs, c1_edge_count, c1_is_ldt, c1_edit_dist = ([] for i in range(4)
) # cograph editing
t1_graphs, t1_edge_count, t1_is_ldt, t1_edit_dist = (
[] for i in range(4)) # triples editing with both insertion/deletion
t2_graphs, t2_edge_count, t2_is_ldt, t2_edit_dist = (
[] for i in range(4)) # triples editing with deletion only
t3_graphs, t3_edge_count, t3_is_ldt, t3_edit_dist = (
[] for i in range(4)) # triples editing with insertion only
t4_graphs, t4_edge_count, t4_is_ldt, t4_edit_dist = ([] for i in range(4)
) # ldt editing
IG1 = None
for i in range(len(species_trees)):
print("Tree pair {}".format(i))
S = PhyloTree.load(name + '/{}'.format(species_trees[i]))
TGT = PhyloTree.load(name + '/{}'.format(gene_trees[i]))
OGT = te.observable_tree(TGT)
ldt = ldt_graph(OGT, S)
if not IG1:
IG1 = InvestigateGraph(ldt)
IG2 = copy.deepcopy(IG1)
IG3 = copy.deepcopy(IG1)
IG4 = copy.deepcopy(IG1)
#IG5 = copy.deepcopy(IG1)
IG1.perturb_graph_terminate(p1, p2)
IG2.perturb_graph_terminate(p1, p2)
IG3.perturb_graph_terminate(p1, p2)
IG4.perturb_graph_terminate(p1, p2)
#IG5.perturb_graph_terminate(p1, p2)
cograph_edited_G, is_c1_ldt, c1_num_edges, c1_ldt_edit_dist = cograph_editing(
IG1)
triples1_edited_G, is_t1_ldt, t1_num_edges, t1_ldt_edit_dist = triples_editing(
IG2, n=100)
triples2_edited_G, is_t2_ldt, t2_num_edges, t2_ldt_edit_dist = triples_editing(
IG3, deletion=True)
triples3_edited_G, is_t3_ldt, t3_num_edges, t3_ldt_edit_dist = triples_editing(
IG4, insertion=True)
#ldt_edited_G, is_t4_ldt, t4_num_edges, t4_ldt_edit_dist = LDT_editing(IG5, deletion = True)
c1_graphs.append(cograph_edited_G)
t1_graphs.append(triples1_edited_G)
t2_graphs.append(triples2_edited_G)
t3_graphs.append(triples3_edited_G)
#t4_graphs.append(ldt_edited_G)
c1_is_ldt.append(is_c1_ldt)
t1_is_ldt.append(is_t1_ldt)
t2_is_ldt.append(is_t2_ldt)
t3_is_ldt.append(is_t3_ldt)
#t4_is_ldt.append(is_t4_ldt)
c1_edge_count.append(c1_num_edges)
t1_edge_count.append(t1_num_edges)
t2_edge_count.append(t2_num_edges)
t3_edge_count.append(t3_num_edges)
#t4_edge_count.append(t4_num_edges)
c1_edit_dist.append(c1_ldt_edit_dist)
t1_edit_dist.append(t1_ldt_edit_dist)
t2_edit_dist.append(t2_ldt_edit_dist)
t3_edit_dist.append(t3_ldt_edit_dist)
#t4_edit_dist.append(t4_ldt_edit_dist)
_, cograph_freq, _ = get_freq(IG1)
triples1_freq, _, _ = get_freq(IG2)
triples2_freq, _, _ = get_freq(IG3)
triples3_freq, _, _ = get_freq(IG4)
#_, _, ldt_freq = get_freq(IG5)
frequencies = [cograph_freq, triples1_freq, triples2_freq, triples3_freq]
return frequencies
def benchmark_fromTrees(n, p1, p2, filename='exact_solutions/trees'):
# load species+gene trees
name = filename + '/{}trees'.format(n)
tree_files = []
#probabilities = [0.15, 0.30, 0.50]
#probs = [(0.15, 0.15), (0.3, 0.3), (0.5, 0.5), (0.15, 0.5), (0.5, 0.15)]
#nodes = [10, 14, 18]
for _, _, files in os.walk(name):
for file in files:
tree_files.append(file)
species_trees = []
gene_trees = []
for f in tree_files:
if f.startswith('species'):
species_trees.append(f)
else:
gene_trees.append(f)
t1_graphs, t1_edge_count, t1_is_ldt, t1_edit_dist = ([] for i in range(4)
) # ldt editing
t2_graphs, t2_edge_count, t2_is_ldt, t2_edit_dist = (
[] for i in range(4)) # ldt editing (triples edit deletion)
t3_graphs, t3_edge_count, t3_is_ldt, t3_edit_dist = (
[] for i in range(4)) # ldt editing (triples edit insertion)
IG1 = None
for i in range(len(species_trees)):
print("Tree pair {}".format(i))
S = PhyloTree.load(name + '/{}'.format(species_trees[i]))
TGT = PhyloTree.load(name + '/{}'.format(gene_trees[i]))
OGT = te.observable_tree(TGT)
ldt = ldt_graph(OGT, S)
if not IG1:
IG1 = InvestigateGraph(ldt)
IG2 = copy.deepcopy(IG1)
IG3 = copy.deepcopy(IG1)
IG1.perturb_graph_terminate(p1, p2)
IG2.perturb_graph_terminate(p1, p2)
IG3.perturb_graph_terminate(p1, p2)
t1_edited_G, is_t1_ldt, _, t1_ldt_edit_dist = LDT_editing(
IG1, n=100
) # ldt editing with triples editing allowing both deletions and insertions for n = 100.
t2_edited_G, is_t2_ldt, _, t2_ldt_edit_dist = LDT_editing(
IG2, deletion=True
) # ldt editing with triples editing allowing only deletions.
t3_edited_G, is_t3_ldt, _, t3_ldt_edit_dist = LDT_editing(
IG3, insertion=True
) # ldt editing with triples editing allowing only insertions.
t1_graphs.append(t1_edited_G)
t2_graphs.append(t2_edited_G)
t3_graphs.append(t3_edited_G)
t1_is_ldt.append(is_t1_ldt)
t2_is_ldt.append(is_t2_ldt)
t3_is_ldt.append(is_t3_ldt)
t1_edit_dist.append(t1_ldt_edit_dist)
t2_edit_dist.append(t2_ldt_edit_dist)
t3_edit_dist.append(t3_ldt_edit_dist)
_, _, ldt1_freq = get_freq(IG1)
_, _, ldt2_freq = get_freq(IG2)
_, _, ldt3_freq = get_freq(IG3)
frequencies = [ldt1_freq, ldt2_freq, ldt3_freq]
return frequencies
def generate_solutions_fromTrees(n, filename):
# load species+gene trees
name = filename + '/{}trees'.format(n)
tree_files = []
probabilities = [0.15, 0.30, 0.50]
#nodes = [10, 14, 18]
restrictions = ['', 'insertion', 'deletion']
for _, _, files in os.walk(name):
for file in files:
tree_files.append(file)
species_trees = []
gene_trees = []
for f in tree_files:
if f.startswith('species'):
species_trees.append(f)
else:
gene_trees.append(f)
ID = 0
for i in range(len(species_trees)):
S = PhyloTree.load(name + '/{}'.format(species_trees[i]))
TGT = PhyloTree.load(name + '/{}'.format(gene_trees[i]))
OGT = te.observable_tree(TGT)
ldt = ldt_graph(OGT, S)
# perturb using p = 0.15, 0.3, 0.5
# for each p, solve using ILP with deletion, insertion and both
for p1 in probabilities:
for p2 in probabilities:
p_i = str(p1).replace('.', '')
if len(p_i) < 3:
p_i = p_i + '0'
p_d = str(p2).replace('.', '')
if len(p_d) < 3:
p_d = p_d + '0'
IG = InvestigateGraph(ldt)
perturbed = IG.perturb_graph_terminate(p1, p2)
if not perturbed:
print("failed")
else:
# solve 3 times using deletion, insertion and both
solver1 = LDTEditor(IG._G_perturbed)
solver1.build_model()
solver1.optimize(time_limit=None)
solver2 = LDTEditor(IG._G_perturbed, only_delete = True)
solver2.build_model()
solver2.optimize(time_limit=None)
solver3 = LDTEditor(IG._G_perturbed, only_add = True)
solver3.build_model()
solver3.optimize(time_limit=None)
sol_graph1, sol_distance1 = solver1.get_solution()
sol_graph2, sol_distance2 = solver2.get_solution()
sol_graph3, sol_distance3 = solver3.get_solution()
properly_colored1 = is_properly_colored(sol_graph1)
cograph1 = is_cograph(sol_graph1)
compatible1 = is_compatible(sol_graph1)
properly_colored2 = is_properly_colored(sol_graph2)
cograph2 = is_cograph(sol_graph2)
compatible2 = is_compatible(sol_graph2)
properly_colored3 = is_properly_colored(sol_graph3)
cograph3 = is_cograph(sol_graph3)
compatible3 = is_compatible(sol_graph3)
folderName = 'exact_solutions/{}_{}_{}nodes{}/'
saveFolder1 = folderName.format(p1, p2, n, '')
if properly_colored1 and cograph1 and compatible1:
print("Saving data...")
solver1._save_ILP_data(IG._G_perturbed, sol_graph1, solver1.get_solve_time(), sol_distance1, i_p = p1, d_p = p2, only_add=False, only_delete=False, saveFolder = folderName.format(p_i, p_d, n, ''), ID = ID)
else:
print("No solution found!")
if properly_colored2 and cograph2 and compatible2:
print("Saving data (deletion)...")
solver2._save_ILP_data(IG._G_perturbed, sol_graph2, solver2.get_solve_time(), sol_distance2, i_p = p1, d_p = p2, only_add=False, only_delete=True, saveFolder = folderName.format(p_i, p_d, n, '_deletion'), ID = ID)
else:
print("No solution found for deletion only!")
if properly_colored3 and cograph3 and compatible3:
print("Saving data (insertion)...")
solver3._save_ILP_data(IG._G_perturbed, sol_graph3, solver3.get_solve_time(), sol_distance3, i_p = p1, d_p = p2, only_add=True, only_delete=False, saveFolder = folderName.format(p_i, p_d, n, '_insertion'), ID = ID)
else:
print("No solution found for insertion only!")
ID += 1
