def BLD_3_Generator(p,q,samplesize,lmbd,C,C2):
#Going to do a fixed species tree with given probabilities of introgression.
outputList=[]
for x in range (0,samplesize):
if numpy.random.random_sample()<p:
outputList.append(expon.rvs()*lmbd)
elif numpy.random.random_sample()<(p+q):
outputList.append(expon.rvs()*lmbd+(C-truncexpon.rvs(C))*lmbd)
else:
outputList.append(expon.rvs()*lmbd+(C2-truncexpon.rvs(C2))*lmbd)
#print (expon.rvs()*lmbd+(C-truncexpon.rvs(C))*lmbd)
return [[('testA','testB','testC'),outputList,[],[]]]
def truncexpon_draws(self, lbd_scale, rndsd=0, thresh=1, datasize=1000):
np.random.seed(rndsd)
for i in range(self.numhyp):
if self.alt_vec[i]==0:
database=truncexpon.rvs(b=thresh, size=datasize)
else:
database=truncexpon.rvs(b=thresh, scale=lbd_scale, size=datasize)
z=sum(database)
pval=1 - norm.cdf(z,loc=datasize*(1+1/(1-np.exp(1))), scale=datasize*(1-np.exp(1)/(np.exp(1)-1)**2))
self.pvec[i]=pval
dirname = './expsettings'
filename = "P_NH%d_PM%.2f_lbd%.2f_SEED%d" % (self.numhyp, self.pi, lbd_scale,rndsd)
saveres(dirname, filename, self.pvec)
def sample_times(node, num_times):
if not hasattr(GC, "NUMPY_SEEDED"):
from numpy.random import seed as numpy_seed
numpy_seed(seed=GC.random_number_seed)
GC.random_number_seed += 1
GC.NUMPY_SEEDED = True
assert hasattr(
GC, 'transmissions'
), "No transmission network found in global context! Run this after the transmission network simulation is done"
first_time = node.get_first_infection_time()
if first_time is None:
return []
windows = []
last_time = first_time
for u, v, t in GC.transmissions:
if u == node and v == node:
if last_time is not None and t > last_time:
windows.append((last_time, t))
last_time = None
elif last_time is None and v == node:
last_time = t
if last_time is not None and t > last_time:
windows.append((last_time, GC.time))
if len(windows) == 0:
windows.append((first_time, GC.time))
truncexpon_variates = (truncexpon.rvs(1, size=num_times))
out = []
for i in range(num_times):
start, end = choice(windows)
out.append((truncexpon_variates[i] * (end - start)) + start)
return out
def _generate_num_repeat(self, scale=2.0):
"""
Helper function, randomly generate number of repetition for an image.
Choose an integer with exponential distribution between [min, max].
Output:
integer between [min_repeat, max_repeat)
"""
min_switch = truncexpon.rvs(b=(self.max_repeat - self.min_repeat) /
scale, loc=self.min_repeat, scale=scale).astype(int)
return min_switch
def exponential(self, N, T):
"""
:return: samples from the exponential distribution with support [self.left, self.right]
Notes - We'll be using exponential with a support 0, 1. To achieve this we'll be
truncating the exponential distribution after 1. The support of the exponential
distribution is from 0 to infinity by default. We will be keeping the scale low so that
not a lot of samples are truncated from the distribution and we have some consistency.
"""
# truncexpon moves from 0 to b.
return truncexpon.rvs(b=self.right,
loc=self.mean,
scale=self.stdev,
size=(N, T),
random_state=self.random_seed)
def day_bucket_to_points(bucket_hist, buckets, kind, n_points=10000):
"""
Transforms one day to points, see `buckets_to_points` docs
:param bucket_hist: One histogram i.e. one day of data from one metric
:param buckets: List of buckets for the histogram
:param kind: Kind of the histogram e.g. 'exponential'
:param n_points: Number of points to be generated. Greater number of points indicates
more accurate point distribution and more time that is needed.
:return: Point representation of one day of data
"""
if kind == 'categorical':
buckets = list(range(len(buckets)))
if kind in ['exponential', 'count']:
points = []
for i, dens in enumerate(bucket_hist):
low = buckets[i]
if i + 1 == len(buckets):
high = low + (buckets[i] - buckets[i - 1])
else:
high = buckets[i + 1]
points += truncexpon.rvs(
high, size=max(int(round(dens*n_points, 0)), 0), loc=low
).tolist()
points = list(points)
else:
points = []
if len(bucket_hist) != len(buckets):
buckets = list(range(len(bucket_hist)))
for i, dens in enumerate(bucket_hist):
low = buckets[i]
if i + 1 == len(buckets):
high = low + (buckets[i] - buckets[i - 1])
else:
high = buckets[i + 1]
points += np.random.uniform(
high=high, low=low,
size=max(int(round(dens*n_points, 0)), 0)
).tolist()
points = list(points)
return points
# Display the probability density function (``pdf``): x = np.linspace(truncexpon.ppf(0.01, b), truncexpon.ppf(0.99, b), 100) ax.plot(x, truncexpon.pdf(x, b), 'r-', lw=5, alpha=0.6, label='truncexpon pdf') # Alternatively, the distribution object can be called (as a function) # to fix the shape, location and scale parameters. This returns a "frozen" # RV object holding the given parameters fixed. # Freeze the distribution and display the frozen ``pdf``: rv = truncexpon(b) ax.plot(x, rv.pdf(x), 'k-', lw=2, label='frozen pdf') # Check accuracy of ``cdf`` and ``ppf``: vals = truncexpon.ppf([0.001, 0.5, 0.999], b) np.allclose([0.001, 0.5, 0.999], truncexpon.cdf(vals, b)) # True # Generate random numbers: r = truncexpon.rvs(b, size=1000) # And compare the histogram: ax.hist(r, normed=True, histtype='stepfilled', alpha=0.2) ax.legend(loc='best', frameon=False) plt.show()
def soc_trunc_filter(traffic, min_soc):
socs = truncexpon.rvs(1, size=traffic, random_state=49816317) # TODO random seed might be counter productive
should_charge = len(socs[socs < min_soc])
return should_charge
def absorption_profile(size): #Exponential random absorption profile
return truncexpon.rvs(thickness / attenuation_length,
scale=attenuation_length,
size=size)
def absorption_profile(size): #Exponential random absorption profile
return truncexpon.rvs(thickness/attenuation_length,
scale=attenuation_length, size=size)
kQ = np.array([
aa_properties[trp]['lambda'] if trp in aa_properties else -1
for trp in tripepsQ
])
idxN = np.argsort(kN)
idxQ = np.argsort(kQ)
tripepsN = tripepsN[idxN]
kN = kN[idxN]
tripepsQ = tripepsQ[idxQ]
kQ = kQ[idxQ]
N = 20000
TmaxN = np.array([-np.log(1 - 0.99) / k for k in kN])
TmaxQ = np.array([-np.log(1 - 0.99) / k for k in kQ])
tN = [truncexpon.rvs(2.5, 0, tm / 2.5, int(N / len(tripepsN))) for tm in TmaxN]
tN = np.concatenate(tN)
tQ = [truncexpon.rvs(2.5, 0, tm / 2.5, int(N / len(tripepsQ))) for tm in TmaxQ]
tQ = np.concatenate(tQ)
sim_lim = 1e-7
deamid_cutoff = 0
# different levels of noise
sigmas = [1, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.15, 0.1, 0.05, 0.005]
bootstraps = 500 # For each level of noise, repeat expetiment
bootstrap_size = 100
corrsN_1 = []
corrsQ_1 = [] # sigmas x bootstraps
corrsN_2 = []
corrsQ_2 = []
corrsN_3 = []
def truncated_exponential_1(b):
return truncexpon.rvs(b=b)
def g_rvs(num_rvs, limit=4.5):
return truncexpon.rvs(loc=limit, b=np.inf, size=num_rvs)
def transmission_neutral_coalescent_tree(transmission_network,
sample_times,
rate=1):
'''Sample a tree under the pure-neutral coalescent model constrained to a given transmission network with given patient sampling times.
Args:
``transmission_network`` (``list``): The transmission network as a ``list`` of ``(u,v,t)`` tuples denoting the transmission from ``u`` to ``v`` at time ``t``. The transmission network must be sorted in ascending order of time
``sample_times`` (``dict``): The times at which each individual was sampled (i.e., the times of the leaves) as a ``dict`` in which keys are individuals from ``transmission_network`` and the value associated with an individual ``u`` is a ``list`` of times at which ``u`` was sampled (i.e., the times of the leaves from individual ``u``)
``rate`` (``float``): The rate of the Poisson process of coalescing two lineages
Returns:
A ``Tree`` object storing the sampled pure-neutral tree, where leaves are labeled ``ID|u|t``, where ``ID`` is a unique identifier for the leaf, ``u`` is the corresponding individual from ``transmission_network``, and ``t`` is the sample time (which equals the leaf's distance from the root). If there are multiple seed individuals (infected beforehand), a tree will be output for each.
'''
if not isinstance(transmission_network, list):
raise TypeError(
"transmission_network must be a list, but it was a %s" %
str(type(transmission_network)))
if not isinstance(sample_times, dict):
raise TypeError("sample_times must be a dict, but it was a %s" %
str(type(sample_times)))
time = dict()
ID = 0
root = dict()
leaves = dict()
infected_by = dict()
to_visit = set()
infection_time = dict()
for i in range(len(transmission_network) - 1, -1, -1):
check_transmission_event(transmission_network[i])
if i != 0 and transmission_network[i][2] < transmission_network[i -
1][2]:
raise ValueError(
"transmission_network must be sorted in ascending order of time"
)
u, v, t = transmission_network[i]
if v in infection_time:
raise ValueError(
"Encountered duplicate transmission recipient: %s" % str(v))
else:
infection_time[v] = t
if u not in infected_by:
infected_by[u] = list()
if v not in infected_by:
infected_by[v] = list()
if v not in sample_times:
raise KeyError("Individual not in sample_times: %s" % str(v))
if not isinstance(sample_times[v], list):
if isinstance(sample_times[v], float) or isinstance(
sample_times[v], int):
sample_times[v] = [sample_times[v]]
elif isinstance(sample_times[v], set):
sample_times[v] = list(sample_times[v])
else:
raise TypeError("Values in sample_times must be list")
if u is not None and u != 'None':
infected_by[u].append(v)
to_visit.add(u)
leaves[v] = list()
for w in infected_by[v]:
if w not in root:
raise ValueError("Missing links in transmission_network")
leaves[v].append(root[w])
del root[w]
for st in sample_times[v]:
if not isinstance(st, float) and not isinstance(st, int):
raise TypeError("Values in sample_times must be list of float")
newnode = Node(label='%d|%s|%f' % (ID, str(v), st))
ID += 1
time[newnode] = st
leaves[v].append(newnode)
leaves[v].sort(key=lambda x: time[x], reverse=True)
lineages = Set()
curr_time = time[leaves[v][0]]
for w in leaves[v]:
if time[w] < curr_time:
while len(lineages) > 1:
L = rate * len(lineages) * (len(lineages) - 1) / 2.
coal_time = curr_time - exponential(1. / L)
if coal_time >= time[w]:
c1 = lineages.pop()
c2 = lineages.pop()
newnode = Node()
time[newnode] = coal_time
curr_time = coal_time
newnode.add_child(c1)
newnode.add_child(c2)
lineages.add(newnode)
curr_time = time[w]
lineages.add(w)
while len(lineages) != 1:
L = rate * len(lineages) * (len(lineages) - 1) / 2.
coal_time = curr_time - truncexpon.rvs(curr_time - t, scale=1. / L)
c1 = lineages.pop()
c2 = lineages.pop()
newnode = Node()
time[newnode] = coal_time
curr_time = coal_time
newnode.add_child(c1)
newnode.add_child(c2)
lineages.add(newnode)
root[v] = lineages.pop()
to_visit.discard(v)
del leaves[v]
if len(to_visit) != 0:
raise ValueError(
"Malformed transmission network. Missing the following seeds:\n%s"
% '\n'.join(str(v) for v in to_visit))
out = list()
for k, r in root.items():
if r.is_root():
tmp = Tree()
tmp.root = r
out.append(tmp)
for node in tmp.traverse_preorder():
if node.is_root():
node.edge_length = time[node] - infection_time[k]
else:
node.edge_length = time[node] - time[node.parent]
return out
aa_properties[trp]['lambda'] if trp in aa_properties else -1
for trp in tripepsQ
])
idxN = np.argsort(kN)
idxQ = np.argsort(kQ)
tripepsN = tripepsN[idxN]
kN = kN[idxN]
tripepsQ = tripepsQ[idxQ]
kQ = kQ[idxQ]
N = 20000
TmaxN = np.array([-np.log(1 - 0.99) / k for k in kN])
TmaxQ = np.array([-np.log(1 - 0.99) / k for k in kQ])
tN = [
truncexpon.rvs(2.35, 0, tm / 2.35, int(N / len(tripepsN))) for tm in TmaxN
]
tN = np.concatenate(tN)
tQ = [truncexpon.rvs(3, 0, tm / 3, int(N / len(tripepsQ))) for tm in TmaxQ]
tQ = np.concatenate(tQ)
sim_lim = 1e-7
deamid_cutoff = 0
s = 0.2
print('Simulating with sigma = {}'.format(s))
sigmasN = np.repeat(s, len(tripepsN))
sigmasQ = np.repeat(s, len(tripepsQ))
# K, sigmas, Tmean, Tstd, Tmax, N, tol
tN_s = tN[np.random.randint(tN.shape[0], size=2000)]
DN = rr.simulate_deamidation(kN, sigmasN, tN_s, sim_lim)
