def _test_survivor_distrib(self):
"""ensure proper number of survivors"""
T = 1.2
birth = 2.0
death = 1.0
ntrees = 10000
survivors = [] # number of survivors in a tree
for i in xrange(ntrees):
tree = birthdeath.sample_birth_death_tree(T, birth, death)
if len(tree.leaves()) == 1:
if tree.leaves()[0].dist >= T:
survivors.append(1)
else:
survivors.append(0)
else:
survivors.append(len(tree.leaves()))
doomprob = birthdeath.prob_birth_death(1, 0, T, birth, death)
h = [(1.0 - doomprob) * x/float(ntrees)
for x in util.hist_int(survivors)]
h[0] = doomprob
h2 = [birthdeath.prob_birth_death(1, x, T, birth, death)
for x in range(0, 10)]
diff = util.vsub(h[:10], h2[:10])
#print h[:10]
#print h2[:10]
self.assert_(max(map(abs, diff)) < .01)
def _test_expected_births(self):
"""ensure proper number of births are occurring"""
T = 1.5
birth = 2.0
death = 0.0
ntrees = 1000
birth_counts = [] # number of births along a random path
for i in xrange(ntrees):
tree = birthdeath.sample_birth_death_tree(T, birth, death)
#treelib.drawTree(tree)
node = tree.root
i = 0
while not node.isLeaf():
node = node.children[random.randint(0, 1)]
i += 1
birth_counts.append(i)
fequal(mean(birth_counts), birth * T, .05)
def test_birth_death_single_sim(self):
"""test the single branch prior"""
duprate = 2.0
lossrate = .5
ntrees = 1000
tabsize = 100
T = 1.0
tops = []
survivors = []
lookup = {}
# define species tree
stree = treelib.parse_newick("(A:1);")
def gene2species(gene):
return gene[:1].upper()
# simulate gene trees
util.tic("simulating %d trees" % ntrees)
for i in xrange(ntrees):
tree, doom = birthdeath.sample_birth_death_tree(
T, duprate, lossrate)
if tree.root in doom:
tops.append("()")
survivors.append(0)
else:
rename_leaves(tree, stree, lambda x: "A")
tops.append(phylo.hash_tree(tree, gene2species))
survivors.append(len(tree.leaves()))
lookup[tops[-1]] = tree
util.toc()
# setup test output
outdir = "test/output/birthdeath_sim_simple"
prep_dir(outdir)
# histogram of topologies and survivors (# leaves)
hist_tops = histtab(tops)
hist_num = histtab(survivors)
# compute survivor prior
probs = []
for row in hist_num:
ngenes = row["item"]
probs.append(birthDeathCount(ngenes, T, duprate, lossrate))
# compute topologie priors
probs_tops = []
for row in hist_tops:
tree = lookup[row["item"]]
if tree.root.is_leaf():
p = log(birthdeath.prob_birth_death1(
0, T, duprate, lossrate))
else:
nhist = numTopologyHistories(tree.root)
s = len(tree.leaves())
thist = factorial(s) * factorial(s-1) / 2**(s-1)
r = numRedunantTopology(tree.root, gene2species,
all_leaves=True)
p = log(r * nhist / thist * birthdeath.prob_birth_death1(
s, T, duprate, lossrate))
probs_tops.append(exp(p))
self.calc_fit(outdir + "/sim_prior_ngenes", hist_num, probs)
self.calc_fit(outdir + "/sim_prior_top", hist_tops, probs_tops)
