def eval_proposal(self, proposal):
"""Compute probability of proposal"""
#util.tic("eval")
# compute recon probability
info = {}
# DEBUG
counts = coal.count_lineages_per_branch(self.coal_tree,
proposal.coal_recon,
proposal.locus_tree)
maxcount = max(x[0] for x in counts.values())
#util.logger("max lineage count %d" % maxcount)
if maxcount > 10:
return -util.INF
p = dlcoal.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,
nsamples=self.nsamples,
add_spec=False,
info=info)
proposal.data = info
#util.toc()
return p
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 assert_bounded_coal_lineages(coal_tree, coal_recon, locus_tree, daughters):
"""Ensure bounded coalescent for daughter lineages (using lineage counts)."""
lineages = coal.count_lineages_per_branch(coal_tree, coal_recon, locus_tree)
for dnode in daughters:
if lineages[dnode][1] != 1:
return False
return True
def assert_bounded_coal_lineages(coal_tree, coal_recon, locus_tree, daughters):
"""Ensure bounded coalescent for daughter lineages (using lineage counts)."""
lineages = coal.count_lineages_per_branch(coal_tree, coal_recon,
locus_tree)
for dnode in daughters:
if lineages[dnode][1] != 1:
return False
return True
def propose_daughters(coal_tree, coal_recon, locus_tree, locus_events):
lineages = coal.count_lineages_per_branch(coal_tree, coal_recon, locus_tree)
daughters = set()
for node, event in locus_events.iteritems():
if event == "dup":
# choose one of the children of node to be a daughter
children = [child for child in node.children
if lineages[child][1] == 1]
if len(children) > 0:
daughters.add(children[stats.sample([1] * len(children))])
return daughters
def propose_daughters(coal_tree, coal_recon, locus_tree, locus_events):
lineages = coal.count_lineages_per_branch(coal_tree, coal_recon,
locus_tree)
daughters = set()
for node, event in locus_events.iteritems():
if event == "dup":
# choose one of the children of node to be a daughter
children = [
child for child in node.children if lineages[child][1] == 1
]
if len(children) > 0:
daughters.add(children[stats.sample([1] * len(children))])
return daughters
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 count_coal(tree, recon, stree):
"""Count the number of (deep) coalescences in a gene tree"""
counts = coal.count_lineages_per_branch(tree, recon, stree)
ncoal = sum(max(x[1] - 1, 0) for x in counts.itervalues())
return ncoal
draw_tree_names(ex["locus_tree"], scale=.5e-7, minlen=8)
print exp(dlcoal.prob_multicoal_recon_topology(
coal_tree, ex["coal_recon"], ex["locus_tree"], n, ex["daughters"]))
print exp(dlcoal.prob_multicoal_recon_topology2(
coal_tree, ex["coal_recon"], ex["locus_tree"], n, ex["daughters"]))
if 0:
draw_tree_names(coal_tree, scale=1e-7)
print exp(coal.prob_multicoal_recon_topology(
coal_tree, ex["coal_recon"], ex["locus_tree"], n))
print exp(coal.prob_multicoal_recon_topology_old(
coal_tree, ex["coal_recon"], ex["locus_tree"], n))
if 0:
pd(coal.count_lineages_per_branch(
coal_tree, ex["coal_recon"], ex["locus_tree"]))
#=============================================================================
show_plots = False
def show_plot():
if show_plots:
raw_input()
if __name__ == "__main__":
if "--" in sys.argv:
def count_coal(tree, recon, stree):
"""Count the number of (deep) coalescences in a gene tree"""
counts = coal.count_lineages_per_branch(tree, recon, stree)
ncoal = sum(max(x[1]-1, 0) for x in counts.itervalues())
return ncoal