def test_regress_mutation_probability_endpoint_conditioning(self):
ngenerations = 10
selection = 2.0
mutation = 0.0001
recombination = 0.001
nchromosomes = 2
npositions = 4
K_initial = np.array([
[1,1,1,1],
[0,0,0,0]], dtype=np.int8)
K_final = np.array([
[1,1,0,0],
[0,0,1,1]], dtype=np.int8)
#
ngenboundaries = ngenerations - 1
no_mutation_prior = (1 - mutation)**(
npositions*ngenboundaries*nchromosomes)
#
initial_long = popgenmarkov.bin2d_to_int(K_initial)
final_long = popgenmarkov.bin2d_to_int(K_final)
ci_to_short, short_to_count, sorted_chrom_lists = get_state_space_info(
nchromosomes, npositions)
initial_ci = chroms_to_index(
sorted(popgenmarkov.bin_to_int(row) for row in K_initial),
npositions)
initial_short = ci_to_short[initial_ci]
final_ci = chroms_to_index(
sorted(popgenmarkov.bin_to_int(row) for row in K_final),
npositions)
final_short = ci_to_short[final_ci]
# get an answer using the less efficient methods
P_sr = popgenmarkov.get_selection_recombination_transition_matrix(
selection, recombination, nchromosomes, npositions)
P_m = popgenmarkov.get_mutation_transition_matrix(
mutation, nchromosomes, npositions)
p_b_given_a = linalg.matrix_power(np.dot(P_sr, P_m), ngenerations-1)[
initial_long, final_long]
p_b_given_a_no_mutation = linalg.matrix_power(P_sr, ngenerations-1)[
initial_long, final_long]
no_mutation_posterior = (
no_mutation_prior * p_b_given_a_no_mutation) / p_b_given_a
# get an answer using the more efficient methods
P_sr_s = get_selection_recombination_transition_matrix_s(
ci_to_short, short_to_count, sorted_chrom_lists,
selection, recombination, nchromosomes, npositions)
P_m_s = get_mutation_transition_matrix_s(
ci_to_short, short_to_count, sorted_chrom_lists,
mutation, nchromosomes, npositions)
p_b_given_a_s = linalg.matrix_power(
np.dot(P_sr_s, P_m_s), ngenerations-1)[
initial_short, final_short]
p_b_given_a_no_mutation_s = linalg.matrix_power(
P_sr_s, ngenerations-1)[
initial_short, final_short]
no_mutation_posterior_s = (
no_mutation_prior * p_b_given_a_no_mutation_s) / p_b_given_a_s
#
self.assertTrue(
np.allclose(no_mutation_posterior, no_mutation_posterior_s))
def test_regress_mutation_probability_endpoint_conditioning(self):
ngenerations = 10
selection = 2.0
mutation = 0.0001
recombination = 0.001
nchromosomes = 2
npositions = 4
K_initial = np.array([[1, 1, 1, 1], [0, 0, 0, 0]], dtype=np.int8)
K_final = np.array([[1, 1, 0, 0], [0, 0, 1, 1]], dtype=np.int8)
#
ngenboundaries = ngenerations - 1
no_mutation_prior = (1 - mutation)**(npositions * ngenboundaries *
nchromosomes)
#
initial_long = popgenmarkov.bin2d_to_int(K_initial)
final_long = popgenmarkov.bin2d_to_int(K_final)
ci_to_short, short_to_count, sorted_chrom_lists = get_state_space_info(
nchromosomes, npositions)
initial_ci = chroms_to_index(
sorted(popgenmarkov.bin_to_int(row) for row in K_initial),
npositions)
initial_short = ci_to_short[initial_ci]
final_ci = chroms_to_index(
sorted(popgenmarkov.bin_to_int(row) for row in K_final),
npositions)
final_short = ci_to_short[final_ci]
# get an answer using the less efficient methods
P_sr = popgenmarkov.get_selection_recombination_transition_matrix(
selection, recombination, nchromosomes, npositions)
P_m = popgenmarkov.get_mutation_transition_matrix(
mutation, nchromosomes, npositions)
p_b_given_a = linalg.matrix_power(np.dot(P_sr, P_m),
ngenerations - 1)[initial_long,
final_long]
p_b_given_a_no_mutation = linalg.matrix_power(
P_sr, ngenerations - 1)[initial_long, final_long]
no_mutation_posterior = (no_mutation_prior *
p_b_given_a_no_mutation) / p_b_given_a
# get an answer using the more efficient methods
P_sr_s = get_selection_recombination_transition_matrix_s(
ci_to_short, short_to_count, sorted_chrom_lists, selection,
recombination, nchromosomes, npositions)
P_m_s = get_mutation_transition_matrix_s(ci_to_short, short_to_count,
sorted_chrom_lists, mutation,
nchromosomes, npositions)
p_b_given_a_s = linalg.matrix_power(np.dot(P_sr_s, P_m_s),
ngenerations - 1)[initial_short,
final_short]
p_b_given_a_no_mutation_s = linalg.matrix_power(
P_sr_s, ngenerations - 1)[initial_short, final_short]
no_mutation_posterior_s = (no_mutation_prior *
p_b_given_a_no_mutation_s) / p_b_given_a_s
#
self.assertTrue(
np.allclose(no_mutation_posterior, no_mutation_posterior_s))
def get_response_content(fs):
initial_state = multiline_state_to_ndarray(fs.initial_state)
final_state = multiline_state_to_ndarray(fs.final_state)
if initial_state.shape != final_state.shape:
raise ValueError(
'initial and final states do not have the same dimensions')
nchromosomes, npositions = initial_state.shape
ndimcap = 12
if nchromosomes * npositions > ndimcap:
raise ValueError('at most 2^%d states are allowed per generation' %
ndimcap)
#
mutation = 0.5 * fs.mutation_param
recombination = 0.5 * fs.recombination_param
selection = fs.selection_param
#
out = StringIO()
print >> out, 'number of chromosomes:'
print >> out, nchromosomes
print >> out
print >> out, 'number of positions per chromosome:'
print >> out, npositions
print >> out
# define the transition matrix
results = pgmfancy.get_state_space_info(nchromosomes, npositions)
ci_to_short, short_to_count, sorted_chrom_lists = results
initial_ci = pgmfancy.chroms_to_index(
sorted(popgenmarkov.bin_to_int(row) for row in initial_state),
npositions)
initial_short = ci_to_short[initial_ci]
final_ci = pgmfancy.chroms_to_index(
sorted(popgenmarkov.bin_to_int(row) for row in final_state),
npositions)
final_short = ci_to_short[final_ci]
P_sr_s = pgmfancy.get_selection_recombination_transition_matrix_s(
ci_to_short, short_to_count, sorted_chrom_lists, selection,
recombination, nchromosomes, npositions)
P_m_s = pgmfancy.get_mutation_transition_matrix_s(ci_to_short,
short_to_count,
sorted_chrom_lists,
mutation, nchromosomes,
npositions)
P = np.dot(P_sr_s, P_m_s)
# sample the endpoint conditioned path
path = PathSampler.sample_endpoint_conditioned_path(
initial_short, final_short, fs.ngenerations, P)
print >> out, 'sampled endpoint conditioned path, including endpoints:'
print >> out
# print the initial state without shuffling of individuals
print >> out, ndarray_to_multiline_state(initial_state)
print >> out
# print the intermediate states with randomly permuted individuals
for short_index in path[1:-1]:
chroms = sorted_chrom_lists[short_index]
random.shuffle(chroms)
K = np.array([popgenmarkov.int_to_bin(x, npositions) for x in chroms])
print >> out, ndarray_to_multiline_state(K)
print >> out
# print the final state without shuffling of individuals
print >> out, ndarray_to_multiline_state(final_state)
print >> out
return out.getvalue()
def get_response_content(fs):
initial_state = multiline_state_to_ndarray(fs.initial_state)
final_state = multiline_state_to_ndarray(fs.final_state)
if initial_state.shape != final_state.shape:
raise ValueError(
'initial and final states do not have the same dimensions')
nchromosomes, npositions = initial_state.shape
if nchromosomes < 2:
raise ValueError(
'please use a population of more than one chromosome')
if npositions < 2:
raise ValueError(
'please use chromosomes longer than one position')
ndimcap = 12
if nchromosomes * npositions > ndimcap:
raise ValueError(
'at most 2^%d states are allowed per generation' % ndimcap)
#
results = pgmfancy.get_state_space_info(nchromosomes, npositions)
ci_to_short, short_to_count, sorted_chrom_lists = results
initial_ci = pgmfancy.chroms_to_index(
sorted(popgenmarkov.bin_to_int(row) for row in initial_state),
npositions)
initial_short = ci_to_short[initial_ci]
final_ci = pgmfancy.chroms_to_index(
sorted(popgenmarkov.bin_to_int(row) for row in final_state),
npositions)
final_short = ci_to_short[final_ci]
#
mutation = fs.mutation_param
recombination = fs.recombination_param
selection = fs.selection_param
# define the prior probabilities of properties of the history
nsiteboundaries = npositions - 1
ngenboundaries = fs.ngenerations - 1
no_mutation_prior = (1 - mutation)**(
npositions*ngenboundaries*nchromosomes)
no_recombination_prior = (1 - recombination)**(
nsiteboundaries*ngenboundaries*nchromosomes)
# get the transition matrices
P_sr_s = pgmfancy.get_selection_recombination_transition_matrix_s(
ci_to_short, short_to_count, sorted_chrom_lists,
selection, recombination, nchromosomes, npositions)
P_s_s = pgmfancy.get_selection_transition_matrix_s(
ci_to_short, short_to_count, sorted_chrom_lists,
selection, nchromosomes, npositions)
P_m_s = pgmfancy.get_mutation_transition_matrix_s(
ci_to_short, short_to_count, sorted_chrom_lists,
mutation, nchromosomes, npositions)
# define some conditional probabilities
p_b_given_a_s = linalg.matrix_power(
np.dot(P_sr_s, P_m_s), fs.ngenerations-1)[
initial_short, final_short]
p_b_given_a_no_mutation_s = linalg.matrix_power(
P_sr_s, fs.ngenerations-1)[
initial_short, final_short]
p_b_given_a_no_recombination_s = linalg.matrix_power(
np.dot(P_s_s, P_m_s), fs.ngenerations-1)[
initial_short, final_short]
# define the conditional properties of properties of the history
no_mutation_posterior = (
no_mutation_prior * p_b_given_a_no_mutation_s) / (
p_b_given_a_s)
no_recombination_posterior = (
no_recombination_prior * p_b_given_a_no_recombination_s) / (
p_b_given_a_s)
#
out = StringIO()
print >> out, 'probability of no mutation in the path history:'
print >> out, 'unconditional: %s' % no_mutation_prior
print >> out, 'endpoint-conditioned: %s' % no_mutation_posterior
print >> out
print >> out, 'probability of no recombination in the path history:'
print >> out, 'unconditional: %s' % no_recombination_prior
print >> out, 'endpoint-conditioned: %s' % no_recombination_posterior
print >> out
return out.getvalue()
def get_response_content(fs):
initial_state = multiline_state_to_ndarray(fs.initial_state)
final_state = multiline_state_to_ndarray(fs.final_state)
if initial_state.shape != final_state.shape:
raise ValueError(
'initial and final states do not have the same dimensions')
nchromosomes, npositions = initial_state.shape
ndimcap = 12
if nchromosomes * npositions > ndimcap:
raise ValueError(
'at most 2^%d states are allowed per generation' % ndimcap)
#
mutation = 0.5 * fs.mutation_param
recombination = 0.5 * fs.recombination_param
selection = fs.selection_param
#
out = StringIO()
print >> out, 'number of chromosomes:'
print >> out, nchromosomes
print >> out
print >> out, 'number of positions per chromosome:'
print >> out, npositions
print >> out
# define the transition matrix
results = pgmfancy.get_state_space_info(nchromosomes, npositions)
ci_to_short, short_to_count, sorted_chrom_lists = results
initial_ci = pgmfancy.chroms_to_index(
sorted(popgenmarkov.bin_to_int(row) for row in initial_state),
npositions)
initial_short = ci_to_short[initial_ci]
final_ci = pgmfancy.chroms_to_index(
sorted(popgenmarkov.bin_to_int(row) for row in final_state),
npositions)
final_short = ci_to_short[final_ci]
P_sr_s = pgmfancy.get_selection_recombination_transition_matrix_s(
ci_to_short, short_to_count, sorted_chrom_lists,
selection, recombination, nchromosomes, npositions)
P_m_s = pgmfancy.get_mutation_transition_matrix_s(
ci_to_short, short_to_count, sorted_chrom_lists,
mutation, nchromosomes, npositions)
P = np.dot(P_sr_s, P_m_s)
# sample the endpoint conditioned path
path = PathSampler.sample_endpoint_conditioned_path(
initial_short, final_short,
fs.ngenerations, P)
print >> out, 'sampled endpoint conditioned path, including endpoints:'
print >> out
# print the initial state without shuffling of individuals
print >> out, ndarray_to_multiline_state(initial_state)
print >> out
# print the intermediate states with randomly permuted individuals
for short_index in path[1:-1]:
chroms = sorted_chrom_lists[short_index]
random.shuffle(chroms)
K = np.array([popgenmarkov.int_to_bin(x, npositions) for x in chroms])
print >> out, ndarray_to_multiline_state(K)
print >> out
# print the final state without shuffling of individuals
print >> out, ndarray_to_multiline_state(final_state)
print >> out
return out.getvalue()
def get_response_content(fs):
initial_state = multiline_state_to_ndarray(fs.initial_state)
final_state = multiline_state_to_ndarray(fs.final_state)
if initial_state.shape != final_state.shape:
raise ValueError(
'initial and final states do not have the same dimensions')
nchromosomes, npositions = initial_state.shape
if nchromosomes < 2:
raise ValueError('please use a population of more than one chromosome')
if npositions < 2:
raise ValueError('please use chromosomes longer than one position')
ndimcap = 12
if nchromosomes * npositions > ndimcap:
raise ValueError('at most 2^%d states are allowed per generation' %
ndimcap)
#
results = pgmfancy.get_state_space_info(nchromosomes, npositions)
ci_to_short, short_to_count, sorted_chrom_lists = results
initial_ci = pgmfancy.chroms_to_index(
sorted(popgenmarkov.bin_to_int(row) for row in initial_state),
npositions)
initial_short = ci_to_short[initial_ci]
final_ci = pgmfancy.chroms_to_index(
sorted(popgenmarkov.bin_to_int(row) for row in final_state),
npositions)
final_short = ci_to_short[final_ci]
#
mutation = fs.mutation_param
recombination = fs.recombination_param
selection = fs.selection_param
# define the prior probabilities of properties of the history
nsiteboundaries = npositions - 1
ngenboundaries = fs.ngenerations - 1
no_mutation_prior = (1 - mutation)**(npositions * ngenboundaries *
nchromosomes)
no_recombination_prior = (1 -
recombination)**(nsiteboundaries *
ngenboundaries * nchromosomes)
# get the transition matrices
P_sr_s = pgmfancy.get_selection_recombination_transition_matrix_s(
ci_to_short, short_to_count, sorted_chrom_lists, selection,
recombination, nchromosomes, npositions)
P_s_s = pgmfancy.get_selection_transition_matrix_s(ci_to_short,
short_to_count,
sorted_chrom_lists,
selection, nchromosomes,
npositions)
P_m_s = pgmfancy.get_mutation_transition_matrix_s(ci_to_short,
short_to_count,
sorted_chrom_lists,
mutation, nchromosomes,
npositions)
# define some conditional probabilities
p_b_given_a_s = linalg.matrix_power(np.dot(P_sr_s, P_m_s),
fs.ngenerations - 1)[initial_short,
final_short]
p_b_given_a_no_mutation_s = linalg.matrix_power(
P_sr_s, fs.ngenerations - 1)[initial_short, final_short]
p_b_given_a_no_recombination_s = linalg.matrix_power(
np.dot(P_s_s, P_m_s), fs.ngenerations - 1)[initial_short, final_short]
# define the conditional properties of properties of the history
no_mutation_posterior = (no_mutation_prior *
p_b_given_a_no_mutation_s) / (p_b_given_a_s)
no_recombination_posterior = (no_recombination_prior *
p_b_given_a_no_recombination_s) / (
p_b_given_a_s)
#
out = StringIO()
print >> out, 'probability of no mutation in the path history:'
print >> out, 'unconditional: %s' % no_mutation_prior
print >> out, 'endpoint-conditioned: %s' % no_mutation_posterior
print >> out
print >> out, 'probability of no recombination in the path history:'
print >> out, 'unconditional: %s' % no_recombination_prior
print >> out, 'endpoint-conditioned: %s' % no_recombination_posterior
print >> out
return out.getvalue()
