def dup_loss_topology_prior(tree, stree, recon, birth, death, maxdoom=20, events=None):
"""
Returns the log prior of a gene tree topology according to dup-loss model
"""
def gene2species(gene):
return recon[tree.nodes[gene]].name
if events is None:
events = phylo.label_events(tree, recon)
leaves = set(tree.leaves())
phylo.add_implied_spec_nodes(tree, stree, recon, events)
pstree, snodes, snodelookup = spidir.make_ptree(stree)
# get doomtable
doomtable = calc_doom_table(stree, birth, death, maxdoom)
prod = 0.0
for node in tree:
if events[node] == "spec":
for schild in recon[node].children:
nodes2 = [x for x in node.children if recon[x] == schild]
if len(nodes2) > 0:
node2 = nodes2[0]
subleaves = get_sub_tree(node2, schild, recon, events)
nhist = birthdeath.num_topology_histories(node2, subleaves)
s = len(subleaves)
thist = stats.factorial(s) * stats.factorial(s - 1) / 2 ** (s - 1)
if len(set(subleaves) & leaves) == 0:
# internal
prod += log(num_redundant_topology(node2, gene2species, subleaves, True))
else:
# leaves
prod += log(num_redundant_topology(node2, gene2species, subleaves, False))
else:
nhist = 1.0
thist = 1.0
s = 0
t = sum(
stats.choose(s + i, i)
* birthdeath.prob_birth_death1(s + i, schild.dist, birth, death)
* exp(doomtable[snodelookup[schild]]) ** i
for i in range(maxdoom + 1)
)
prod += log(nhist) - log(thist) + log(t)
# correct for renumbering
nt = num_redundant_topology(tree.root, gene2species)
prod -= log(nt)
# phylo.removeImpliedSpecNodes(tree, recon, events)
treelib.remove_single_children(tree)
return prod
def count_dup_loss_coal_tree(coal_tree, extra, stree, gene2species,
implied=True, locus_mpr=True):
"""count dup loss coal"""
if not locus_mpr:
raise Exception("not implemented")
# TODO: use locus_recon and locus_events rather than MPR
# (currently, phylo.py reconciliation functions fail for non-MPR)
locus_tree = extra["locus_tree"]
locus_recon = phylo.reconcile(locus_tree, stree, gene2species)
locus_events = phylo.label_events(locus_tree, locus_recon)
coal_recon = extra["coal_recon"]
ndup, nloss, nappear = phylo.count_dup_loss_tree(locus_tree, stree, gene2species,
locus_recon, locus_events)
# add implied speciation nodes if desired
# this must be added AFTER counting dups and losses since it affects loss inference
if implied:
added = phylo.add_implied_spec_nodes(locus_tree, stree, locus_recon, locus_events)
# count coals
ncoal = 0
counts = coal.count_lineages_per_branch(coal_tree, coal_recon, locus_tree)
for lnode, (count_bot, count_top) in counts.iteritems():
n = max(count_top-1, 0)
locus_recon[lnode].data['coal'] += n
ncoal += n
if implied:
phylo.remove_implied_spec_nodes(locus_tree, added, locus_recon, locus_events)
return ndup, nloss, ncoal, nappear
def eval_proposal(self, proposal):
"""Compute cost of proposal"""
if not phyloDLC.assert_daughters(proposal.locus_events, proposal.daughters):
# ensure locus events (duplications) and daughters match
ndup, nloss, ncoal = None, None, None
dupcost, losscost, coalcost = util.INF, util.INF, util.INF
else:
# find dup cost
if self.dupcost == 0:
ndup = None
dupcost = 0
else:
ndup = phylo.count_dup(proposal.locus_tree, proposal.locus_events)
dupcost = ndup * self.dupcost
# find loss cost
if self.losscost == 0:
nloss = None
losscost = 0
else:
nloss = phylo.count_loss(proposal.locus_tree, self.stree, proposal.locus_recon)
losscost = nloss * self.losscost
# find coal cost (first ensure bounded coalescent is satisfied - should always be true based on how daughters are proposed)
phyloDLC.assert_bounded_coal(self.coal_tree, proposal.coal_recon, proposal.locus_tree, proposal.daughters)
if self.coalcost == 0:
ncoal = None
coalcost = 0
else:
# add implied speciation nodes if desired
# this must be added AFTER counting dups and losses since it affects loss inference
if self.implied:
added = phylo.add_implied_spec_nodes(proposal.locus_tree, self.stree,
proposal.locus_recon, proposal.locus_events)
ncoal = phyloDLC.count_coal(self.coal_tree, proposal.coal_recon, proposal.locus_tree)
coalcost = ncoal * self.coalcost
if self.implied:
phylo.remove_implied_spec_nodes(proposal.locus_tree, added,
proposal.locus_recon, proposal.locus_events)
# total cost
cost = dupcost + losscost + coalcost
# logging info
info = {}
info["ndup"] = ndup
info["nloss"] = nloss
info["ncoal"] = ncoal
info["cost"] = cost
proposal.data = info
return cost
def eval_proposal(self, proposal):
"""Compute probability of proposal"""
# compute recon probability
phylo.add_implied_spec_nodes(proposal["locus_tree"], self.stree,
proposal["locus_recon"],
proposal["locus_events"])
p = prob_dlcoal_recon_topology(self.coal_tree,
proposal["coal_recon"],
proposal["locus_tree"],
proposal["locus_recon"],
proposal["locus_events"],
proposal["daughters"],
self.stree, self.n,
self.duprate, self.lossrate,
self.pretime, self.premean,
maxdoom=self.maxdoom,
nsamples=self.nsamples,
add_spec=False)
treelib.remove_single_children(proposal["locus_tree"])
phylo.subset_recon(proposal["locus_tree"], proposal["locus_recon"])
return p
def add_implied_spec_nodes(tree, stree, recon, events):
"""
Add speciation nodes to tree that are implied but are not present because of gene losses.
Extends phylo.add_implied_spec_nodes to handle non-MPR.
Only guaranteed to work for binary trees.
"""
added_spec = phylo.add_implied_spec_nodes(tree, stree, recon, events)
added_dup = []
for node in list(tree):
schildren = [recon[child] for child in node.children]
if len(schildren) > 1 and len(set(schildren)) == 1 and events[node] != "dup":
added_dup.extend(add_spec_from_dup_nodes(node, tree, recon, events))
assert is_full_tree(tree, stree, recon, events)
return added_spec, added_dup
def count_dup_loss_coal(coal_tree, extra, stree, implied=True):
"""Returns the number of duplications + transfers + losses in a gene tree"""
locus_tree = extra["locus_tree"]
locus_recon = extra["locus_recon"]
locus_events = extra["locus_events"]
coal_recon = extra["coal_recon"]
ndup = count_dup(locus_tree, locus_events)
nloss = count_loss(locus_tree, stree, locus_recon)
if implied:
# add implied speciation nodes if desired
# this must be added AFTER counting dups and losses since it affects loss inference
added = phylo.add_implied_spec_nodes(locus_tree, stree, locus_recon, locus_events)
ncoal = count_coal(coal_tree, coal_recon, locus_tree)
if implied:
phylo.remove_implied_spec_nodes(locus_tree, added, locus_recon, locus_events)
return ndup + nloss + ncoal
def count_dup_loss_coal_tree(coal_tree,
extra,
stree,
gene2species,
implied=True,
locus_mpr=True):
"""count dup loss coal"""
if not locus_mpr:
raise Exception("not implemented")
# TODO: use locus_recon and locus_events rather than MPR
# (currently, phylo.py reconciliation functions fail for non-MPR)
locus_tree = extra["locus_tree"]
locus_recon = phylo.reconcile(locus_tree, stree, gene2species)
locus_events = phylo.label_events(locus_tree, locus_recon)
coal_recon = extra["coal_recon"]
ndup, nloss, nappear = phylo.count_dup_loss_tree(locus_tree, stree,
gene2species, locus_recon,
locus_events)
# add implied speciation nodes if desired
# this must be added AFTER counting dups and losses since it affects loss inference
if implied:
added = phylo.add_implied_spec_nodes(locus_tree, stree, locus_recon,
locus_events)
# count coals
ncoal = 0
counts = coal.count_lineages_per_branch(coal_tree, coal_recon, locus_tree)
for lnode, (count_bot, count_top) in counts.iteritems():
n = max(count_top - 1, 0)
locus_recon[lnode].data['coal'] += n
ncoal += n
if implied:
phylo.remove_implied_spec_nodes(locus_tree, added, locus_recon,
locus_events)
return ndup, nloss, ncoal, nappear
def add_implied_spec_nodes(tree, stree, recon, events):
"""
Add speciation nodes to tree that are implied but are not present because of gene losses.
Extends phylo.add_implied_spec_nodes to handle non-MPR.
Only guaranteed to work for binary trees.
"""
added_spec = phylo.add_implied_spec_nodes(tree, stree, recon, events)
added_dup = []
for node in list(tree):
schildren = [recon[child] for child in node.children]
if len(schildren) > 1 and len(
set(schildren)) == 1 and events[node] != "dup":
added_dup.extend(add_spec_from_dup_nodes(node, tree, recon,
events))
assert is_full_tree(tree, stree, recon, events)
return added_spec, added_dup
def count_dup_loss_coal(coal_tree, extra, stree, implied=True):
"""Returns the number of duplications + transfers + losses in a gene tree"""
locus_tree = extra["locus_tree"]
locus_recon = extra["locus_recon"]
locus_events = extra["locus_events"]
coal_recon = extra["coal_recon"]
ndup = count_dup(locus_tree, locus_events)
nloss = count_loss(locus_tree, stree, locus_recon)
if implied:
# add implied speciation nodes if desired
# this must be added AFTER counting dups and losses since it affects loss inference
added = phylo.add_implied_spec_nodes(locus_tree, stree, locus_recon,
locus_events)
ncoal = count_coal(coal_tree, coal_recon, locus_tree)
if implied:
phylo.remove_implied_spec_nodes(locus_tree, added, locus_recon,
locus_events)
return ndup + nloss + ncoal
def eval_proposal(self, proposal):
"""Compute cost of proposal"""
if not phyloDLC.assert_daughters(proposal.locus_events,
proposal.daughters):
# ensure locus events (duplications) and daughters match
ndup, nloss, ncoal = None, None, None
dupcost, losscost, coalcost = util.INF, util.INF, util.INF
else:
# find dup cost
if self.dupcost == 0:
ndup = None
dupcost = 0
else:
ndup = phylo.count_dup(proposal.locus_tree,
proposal.locus_events)
dupcost = ndup * self.dupcost
# find loss cost
if self.losscost == 0:
nloss = None
losscost = 0
else:
nloss = phylo.count_loss(proposal.locus_tree, self.stree,
proposal.locus_recon)
losscost = nloss * self.losscost
# find coal cost (first ensure bounded coalescent is satisfied - should always be true based on how daughters are proposed)
phyloDLC.assert_bounded_coal(self.coal_tree, proposal.coal_recon,
proposal.locus_tree,
proposal.daughters)
if self.coalcost == 0:
ncoal = None
coalcost = 0
else:
# add implied speciation nodes if desired
# this must be added AFTER counting dups and losses since it affects loss inference
if self.implied:
added = phylo.add_implied_spec_nodes(
proposal.locus_tree, self.stree, proposal.locus_recon,
proposal.locus_events)
ncoal = phyloDLC.count_coal(self.coal_tree,
proposal.coal_recon,
proposal.locus_tree)
coalcost = ncoal * self.coalcost
if self.implied:
phylo.remove_implied_spec_nodes(proposal.locus_tree, added,
proposal.locus_recon,
proposal.locus_events)
# total cost
cost = dupcost + losscost + coalcost
# logging info
info = {}
info["ndup"] = ndup
info["nloss"] = nloss
info["ncoal"] = ncoal
info["cost"] = cost
proposal.data = info
return cost
def dlcoal_recon_old(tree, stree, gene2species,
n, duprate, lossrate,
pretime=None, premean=None,
nsearch=1000,
maxdoom=20, nsamples=100,
search=phylo.TreeSearchNni):
"""
Perform reconciliation using the DLCoal model
Returns (maxp, maxrecon) where 'maxp' is the probability of the
MAP reconciliation 'maxrecon' which further defined as
maxrecon = {'coal_recon': coal_recon,
'locus_tree': locus_tree,
'locus_recon': locus_recon,
'locus_events': locus_events,
'daughters': daughters}
"""
# init coal tree
coal_tree = tree
# init locus tree as congruent to coal tree
# equivalent to assuming no ILS
locus_tree = coal_tree.copy()
maxp = - util.INF
maxrecon = None
# init search
locus_search = search(locus_tree)
for i in xrange(nsearch):
# TODO: propose other reconciliations beside LCA
locus_tree2 = locus_tree.copy()
phylo.recon_root(locus_tree2, stree, gene2species, newCopy=False)
locus_recon = phylo.reconcile(locus_tree2, stree, gene2species)
locus_events = phylo.label_events(locus_tree2, locus_recon)
# propose daughters (TODO)
daughters = set()
# propose coal recon (TODO: propose others beside LCA)
coal_recon = phylo.reconcile(coal_tree, locus_tree2, lambda x: x)
# compute recon probability
phylo.add_implied_spec_nodes(locus_tree2, stree,
locus_recon, locus_events)
p = prob_dlcoal_recon_topology(coal_tree, coal_recon,
locus_tree2, locus_recon, locus_events,
daughters,
stree, n, duprate, lossrate,
pretime, premean,
maxdoom=maxdoom, nsamples=nsamples,
add_spec=False)
treelib.remove_single_children(locus_tree2)
if p > maxp:
maxp = p
maxrecon = {"coal_recon": coal_recon,
"locus_tree": locus_tree2,
"locus_recon": locus_recon,
"locus_events": locus_events,
"daughters": daughters}
locus_tree = locus_tree2.copy()
locus_search.set_tree(locus_tree)
else:
locus_search.revert()
# perform local rearrangement to locus tree
locus_search.propose()
return maxp, maxrecon
def prob_dlcoal_recon_topology(coal_tree, coal_recon,
locus_tree, locus_recon, locus_events,
daughters,
stree, n, duprate, lossrate,
pretime=None, premean=None,
maxdoom=20, nsamples=100,
add_spec=True):
"""
Probability of a reconcile gene tree in the DLCoal model.
coal_tree -- coalescent tree
coal_recon -- reconciliation of coalescent tree to locus tree
locus_tree -- locus tree (has dup-loss)
locus_recon -- reconciliation of locus tree to species tree
locus_events -- events dict for locus tree
stree -- species tree
n -- population sizes in species tree
duprate -- duplication rate
lossrate -- loss rate
You must also specify one of the following
pretime -- starting time before species tree
premean -- mean starting time before species tree
Note: locus tree must have implied speciation nodes present
"""
dups = phylo.count_dup(locus_tree, locus_events)
# ensure implicit speciations are present
if add_spec:
phylo.add_implied_spec_nodes(locus_tree, stree,
locus_recon, locus_events)
# init popsizes for locus tree
stree_popsizes = coal.init_popsizes(stree, n)
popsizes = {}
for node in locus_tree:
popsizes[node.name] = stree_popsizes[locus_recon[node].name]
# duploss probability
util.tic("top")
dl_prob = spidir.calc_birth_death_prior(locus_tree, stree, locus_recon,
duprate, lossrate,
maxdoom=maxdoom)
util.toc()
# daughters probability
d_prob = dups * log(.5)
# integrate over duplication times using sampling
prob = 0.0
#util.tic("int")
for i in xrange(nsamples):
# sample duplication times
locus_times = spidir.topology_prior.sample_dup_times(
locus_tree, stree, locus_recon, duprate, lossrate, pretime,
premean,
events=locus_events)
assert len(locus_times) == len(locus_tree.nodes), (
len(locus_times), len(locus_tree.nodes))
treelib.set_dists_from_timestamps(locus_tree, locus_times)
# coal topology probability
coal_prob = prob_coal_recon_topology(coal_tree, coal_recon,
locus_tree, popsizes, daughters)
prob += exp(coal_prob)
print coal_prob
#util.toc()
return dl_prob + d_prob + util.safelog(prob / nsamples)
def dup_loss_topology_prior(tree,
stree,
recon,
birth,
death,
maxdoom=20,
events=None):
"""
Returns the log prior of a gene tree topology according to dup-loss model
"""
def gene2species(gene):
return recon[tree.nodes[gene]].name
if events is None:
events = phylo.label_events(tree, recon)
leaves = set(tree.leaves())
phylo.add_implied_spec_nodes(tree, stree, recon, events)
pstree, snodes, snodelookup = spidir.make_ptree(stree)
# get doomtable
doomtable = calc_doom_table(stree, birth, death, maxdoom)
prod = 0.0
for node in tree:
if events[node] == "spec":
for schild in recon[node].children:
nodes2 = [x for x in node.children if recon[x] == schild]
if len(nodes2) > 0:
node2 = nodes2[0]
subleaves = get_sub_tree(node2, schild, recon, events)
nhist = birthdeath.num_topology_histories(node2, subleaves)
s = len(subleaves)
thist = stats.factorial(s) * stats.factorial(s - 1) / 2**(
s - 1)
if len(set(subleaves) & leaves) == 0:
# internal
prod += log(
num_redundant_topology(node2, gene2species,
subleaves, True))
else:
# leaves
prod += log(
num_redundant_topology(node2, gene2species,
subleaves, False))
else:
nhist = 1.0
thist = 1.0
s = 0
t = sum(
stats.choose(s + i, i) * birthdeath.prob_birth_death1(
s + i, schild.dist, birth, death) *
exp(doomtable[snodelookup[schild]])**i
for i in range(maxdoom + 1))
prod += log(nhist) - log(thist) + log(t)
# correct for renumbering
nt = num_redundant_topology(tree.root, gene2species)
prod -= log(nt)
#phylo.removeImpliedSpecNodes(tree, recon, events)
treelib.remove_single_children(tree)
return prod