def test_CompareUsingView(self):
"""
Evolve some population, take samples,
get details of those samples, make sure
those details match up what 'views' think should
be in the population.
"""
samples = [fp.get_samples(rng,i,100) for i in pops]
details = [fp.get_sample_details(i[1],j) for i,j in zip(samples,pops)]
##For each element in details,
##find it in the "mutation view" for that replicate
for i in range(len(samples)):
##Make sure that lengths match up
for key in details[i]:
self.assertEqual(len(samples[i][1]),len(details[i][key]))
for j in range(len(samples[i][1])):
mm=[X for X in mviews[i] if X['pos'] == samples[i][1][j][0]]
##Make sure each position is uniuqe
self.assertEqual(len(mm),1)
for mmi in mm:
##Make sure that the position is equal to what we expect
self.assertEqual(mmi['pos'],samples[i][1][j][0])
##Make sure selection coefficient matches up
self.assertEqual(mmi['s'],details[i]['s'][j])
EP=mmi['n']/float(2000) #"expected" frequency
self.assertEqual(EP,details[i]['p'][j])
def test_CountFixationsInSample2(self):
#Repeat above test, but using sampler where neutral + selected variants are in different objects
samples = [fp.get_samples(rng,i,100,False) for i in pops]
for i in range(len(fixations)):
FOUNDN=0
FOUNDS=0
for f in fixations[i]:
if f['neutral'] is True:
mm=[X for X in samples[i][0] if X[0]==f['pos']]
FOUNDN+=len(mm)
else:
mm=[X for X in samples[i][1] if X[0]==f['pos']]
FOUNDS+=len(mm)
self.assertEqual(FOUNDN+FOUNDS,len(fixations[i]))
def test_CountFixationsInSample3(self):
#Finally, make sure that sample w/o including fixations works,
#meaning that no mutations are present in the sample with
#derived count == nsam
samples = [fp.get_samples(rng,i,100) for i in pops]
##Checks that no "true fixations" are in vector
for i in range(len(fixations)):
FOUNDN=0
FOUNDS=0
for f in fixations[i]:
if f['neutral'] is True:
mm=[X for X in samples[i][0] if X[0]==f['pos']]
FOUNDN+=len(mm)
else:
mm=[X for X in samples[i][1] if X[0]==f['pos']]
FOUNDS+=len(mm)
self.assertEqual(FOUNDN+FOUNDS,0)
##Checks that no polymorphisms are present as fixations in sample
for i in samples:
for j in i:
x=j[0][1].count(b'1')
self.assertTrue(x<100)
x=j[1][1].count(b'1')
self.assertTrue(x<100)
# In[8]:
#Now, pops is a Python list with len(pops) = 40
#Each element's type is fwdpy.singlepop
print(len(pops))
print(type(pops[0]))
# ## Taking samples from simulated populations
# In[9]:
#Use a list comprehension to get a random sample of size
#n = 20 from each replicate
samples = [fp.get_samples(rng,i,20) for i in pops]
#Samples is now a list of tuples of two lists.
#Each list contains tuples of mutation positions and genotypes.
#The first list represents neutral variants.
#The second list represents variants affecting fitness ('selected' variants)
#We will manipulate/analyze these genotypes, etc.,
#in a later example
for i in samples[:4]:
print ("A sample from a population is a ",type(i))
print(len(samples))
# ### Getting additional information about samples
import fwdpy
import pandas
rng = fwdpy.GSLrng(100)
pop = fwdpy.evolve_pops_t(rng,3,1000,[1000]*int(1e4),50,50)
s = [fwdpy.get_samples(rng,i,[10,]) for i in pop]
###fxn to ask if a site is a singleton
def isSingleton( site ):
ones=site[1].count('1')
if ones == 1:
return True
return False
##fxn to count derived singletonss in a sample
def countDerived( sample ):
nsing=0
for i in range(len(sample)):
if isSingleton(sample[i]):
nsing += 1
return nsing
#list comprehension for automatic vectorizing
D = [fwdpy.TajimasD(si[0]) for si in s]
print "Tajima's D per sample =", D
#number of seg sites per sample
segsites = [len(si[0]) for si in s]
print "S per sample =",segsites
#number of singletons per sample
#The first part is the same as the example for fwdpy.evolve_regions import fwdpy import numpy as np nregions = [fwdpy.Region(0,1,1),fwdpy.Region(2,3,1)] sregions = [fwdpy.ExpS(1,2,1,-0.001,0.0),fwdpy.ExpS(1,2,0.01,0.001)] rregions = [fwdpy.Region(0,3,1)] rng = fwdpy.GSLrng(100) popsizes = np.array([1000],dtype=np.uint32) # Evolve for 5N generations initially popsizes=np.tile(popsizes,10000) pops = fwdpy.evolve_regions(rng,1,1000,popsizes[0:],0.001,0.01,0.001,nregions,sregions,rregions) #Now, "bud" off a daughter population of same size, and evolve both for another 100 generations mpops = fwdpy.evolve_regions_split(rng,pops,popsizes[0:100],popsizes[0:100],0.001,0.0001,0.001,nregions,sregions,rregions) samples = [fwdpy.get_samples(rng,i,[10,10]) for i in mpops] details = [ fwdpy.get_sample_details(rng,i[1],j)
)
# evolve_regions() returns a 'popvec' object. It is an iterablet hat contains
# a bunch of 'singlepop' objects.
# Let's get some basic information about each one
for index, p in enumerate(pops):
print "Population", index
print " Evolved to generation", p.gen()
print " Has", p.popsize(), "diploid individuals"
# Now, to calculate some summary statistics
# fwdpy has a handle into the libsequence library
import fwdpy.libseq
# To calculate summary stats, we have to take a sample from our populations
# get_samples(
# RNG,
# singlepop object,
# nsam,
# removeFixed=True, (True: keep only polymorphic sites)
# deme=None (required for if there is pop structure)
# )
samples = [fwdpy.get_samples(rng, p, 20) for p in pops]
# Then we can calculate windowed summary statistics
# windows(
windowed_stats = [
fwdpp.libseq.windows(i[0], 0.1, 0.1, -2, 2) for i in samples
]