def sc3ei(params , ns):
# p_misid: proportion of misidentified ancestral states
# P: proportion of sites with lower migration
# Fi: factor of migration reduction (1e-5 - 0.999)
# migration is asymmetric, scales with pop size
nu1_1,nu2_1,nu1_2,nu2_2,T1,T2,m12,m21,Fs,P,p_misid = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1_1, nu2_1], T1, m = np.array([[0, m12], [m21, 0]]))
fs.integrate([nu1_2, nu2_2], T2, m = np.array([[0, 0], [0, 0]]))
stsi = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsi = moments.Spectrum(stsi)
fsi = moments.Manips.split_1D_to_2D(fsi, ns[0], ns[1])
fsi.integrate([nu1_1, nu2_1*Fs], T1, m = np.array([[0, m12*Fs], [m21*Fs, 0]]))
fsi.integrate([nu1_2, nu2_2*Fs], T2, m = np.array([[0, 0], [0, 0]]))
fs2=P*fsi+(1-P)*fs
return (1-p_misid)*fs2 + p_misid*moments.Numerics.reverse_array(fs2)
def SC(params, ns):
s, nToro, nMega, t1, t2, m12, m21 = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
nu1_func = lambda t: s * (nToro / s)**(t / t2)
nu2_func = lambda t: (1 - s) * (nMega / (1 - s))**(t / t2)
nu_func = lambda t: [nu1_func(t), nu2_func(t)]
fs.integrate([nToro, nMega], t1, dt_fac=0.01, m=np.array([[0, 0], [0, 0]]))
fs.integrate(nu_func, t2, dt_fac=0.01, m=np.array([[0, m12], [m21, 0]]))
return fs
def sc11(params, ns):
# p_misid: proportion of misidentified ancestral states
nu0, nu1_2, nu2_2, T0, T, m = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs.integrate([nu0], T0)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1_2, nu2_2], T, m=np.array([[0, m], [m, 0]]))
return fs
def model_func(params, ns): t1, nu11, t2, nu21, s1, t3, nu31, nu32, m3_12, m3_21 = params sts = moments.LinearSystem_1D.steady_state_1D(np.sum(ns)) fs = moments.Spectrum(sts) fs.integrate(tf=t1, Npop=[nu11], dt_fac=0.01) fs.integrate(tf=t2, Npop=[nu21], dt_fac=0.01) fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1]) nu1_func = lambda t: (s1 * nu21) * (nu31 / (s1 * nu21)) ** (t / t3) migs = np.array([[0, m3_12], [m3_21, 0]]) fs.integrate(tf=t3, Npop=lambda t: [nu1_func(t), nu32], m=migs, dt_fac=0.01) return fs
def sc3ei(params, ns):
# p_misid: proportion of misidentified ancestral states
nu1_1, nu2_1, nu1_2, nu2_2, nu1_3, nu2_3, T1, T2, T3, m12_3, m21_3 = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1_1, nu2_1], T1, m=np.array([[0, m12_3], [m21_3, 0]]))
fs.integrate([nu1_2, nu2_2], T2, m=np.array([[0, 0], [0, 0]]))
fs.integrate([nu1_3, nu2_3], T3, m=np.array([[0, 0], [0, 0]]))
return fs
def sc12nm(params, ns):
# p_misid: proportion of misidentified ancestral states
nu0, nu1, T1, T2 = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs.integrate([nu0], T1)
fs.integrate([nu1], T2)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
return fs
def test_expected_sfs(self):
pop_id = 'root'
ns0 = 20
nu = 1.
theta = 1.0
fs_moments = moments.Spectrum(
moments.LinearSystem_1D.steady_state_1D(ns0, gamma=0, h=0.5),
pop_ids=[pop_id])
fs = demography.integration.moments_fs_root_equilibrium(ns0, nu, theta, pop_id)
self.assertTrue(np.allclose(fs_moments, fs))
self.assertTrue(fs.pop_ids == [pop_id])
def sc11(params, ns):
# p_misid: proportion of misidentified ancestral states
nu0, nu1_2, nu2_2, T0, T, m12_2, m21_2, p_misid = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs.integrate([nu0], T0)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1_2, nu2_2], T, m=np.array([[0, m12_2], [m21_2, 0]]))
return (1 - p_misid) * fs + p_misid * moments.Numerics.reverse_array(fs)
def test_marginalize(self):
ns = (7, 8, 6)
fs = moments.Spectrum(numpy.random.uniform(size=ns))
folded = fs.fold()
marg1 = fs.marginalize([1])
# Do manual marginalization.
manual = moments.Spectrum(fs.data.sum(axis=1))
# Check that these are equal in the unmasked entries.
self.assert_(
numpy.allclose(numpy.where(marg1.mask, 0, marg1.data),
numpy.where(manual.mask, 0, manual.data)))
# Check folded Spectrum objects. I should get the same result if I
# marginalize then fold, as if I fold then marginalize.
mf1 = marg1.fold()
mf2 = folded.marginalize([1])
self.assert_(numpy.allclose(mf1, mf2))
def sc12iml(params, ns):
# p_misid: proportion of misidentified ancestral states
nu0, nu1_2, nu2_2, nu1_3, nu2_3, T1, T2, T3, m, mi, m12_3, m21_3, m12_3i, m21_3i, P = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs.integrate([nu0], T1)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1_2, nu2_2], T2, m=np.array([[0, m], [m, 0]]))
fs.integrate([nu1_3, nu2_3], T3, m=np.array([[0, m12_3], [m21_3, 0]]))
stsi = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsi = moments.Spectrum(stsi)
fsi.integrate([nu0], T1)
fsi = moments.Manips.split_1D_to_2D(fsi, ns[0], ns[1])
fsi.integrate([nu1_2, nu2_2], T2, m=np.array([[0, mi], [mi, 0]]))
fsi.integrate([nu1_3, nu2_3], T3, m=np.array([[0, m12_3i], [m21_3i, 0]]))
fs2 = P * fsi + (1 - P) * fs
return fs2
def sc12nm(params , ns):
# p_misid: proportion of misidentified ancestral states
# P: proportion of sites with lower Ne
# Fs: factor of Ne reduction (1e-5 - 0.999)
nu0,nu1,T1,T2,Fs,P = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs.integrate([nu0], T1)
fs.integrate([nu1], T2)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
stsi = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsi = moments.Spectrum(sts)
fsi.integrate([nu0*Fs], T1)
fsi.integrate([nu1*Fs], T2)
fsi = moments.Manips.split_1D_to_2D(fsi, ns[0], ns[1])
fs2=P*fsi+(1-P)*fs
return fs2
def sc3ei(params, ns):
# p_misid: proportion of misidentified ancestral states
# Pi: proportion of sites with lower migration
# Fi: factor of migration reduction (1e-5 - 0.999)
# Ps: proportion of sites with lower Ne
# Fs: factor of Ne reduction (0.1 - 0.999)
# migration is asymmetric, scales with pop size
nu1_1, nu2_1, nu1_2, nu2_2, T1, T2, m12, m21, Fi, Pi, Fs, Ps = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1_1, nu2_1], T1, m=np.array([[0, m12], [m21, 0]]))
fs.integrate([nu1_2, nu2_2], T2, m=np.array([[0, 0], [0, 0]]))
stsi = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsi = moments.Spectrum(stsi)
fsi = moments.Manips.split_1D_to_2D(fsi, ns[0], ns[1])
fsi.integrate([nu1_1, nu2_1],
T1,
m=np.array([[0, m12 * Fi], [m21 * Fi, 0]]))
fsi.integrate([nu1_2, nu2_2], T2, m=np.array([[0, 0], [0, 0]]))
stsis = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsis = moments.Spectrum(stsis)
fsis = moments.Manips.split_1D_to_2D(fsis, ns[0], ns[1])
fsis.integrate([nu1_1 * Fs, nu2_1 * Fs],
T1,
m=np.array([[0, m12 * Fi * Fs], [m21 * Fi * Fs, 0]]))
fsis.integrate([nu1_2 * Fs, nu2_2 * Fs], T2, m=np.array([[0, 0], [0, 0]]))
stss = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fss = moments.Spectrum(stss)
fss = moments.Manips.split_1D_to_2D(fss, ns[0], ns[1])
fss.integrate([nu1_1 * Fs, nu2_1 * Fs],
T1,
m=np.array([[0, m12 * Fs], [m21 * Fs, 0]]))
fss.integrate([nu1_2 * Fs, nu2_2 * Fs], T2, m=np.array([[0, 0], [0, 0]]))
fs2 = Pi * (1 - Ps) * fsi + Ps * (1 - Pi) * fss + Pi * Ps * fsis + (
1 - Pi) * (1 - Ps) * fs
return fs2
def test_pickle(self):
"""
Saving spectrum to file.
"""
comments = ['comment 1', 'comment 2']
filename = 'test.p'
data = numpy.random.rand(3, 3)
fs = moments.Spectrum(data)
pickle.dump(fs, open(filename, "wb"))
os.remove(filename)
def s12nm(params , ns):
# p_misid: proportion of misidentified ancestral states
# P: proportion of sites with lower Ne
# Fs: factor of Ne reduction (1e-5 - 0.999)
nu1,nu1_2,nu2_2,T1,T2 = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs.integrate([nu1], T1)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1_2, nu2_2], T2, m = np.array([[0, 0], [0, 0]]))
return fs
def sc12nm(params, ns):
# p_misid: proportion of misidentified ancestral states
nu0, nu1, nu2, T1, T2, T3, p_misid = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs.integrate([nu0], T1)
fs.integrate([nu1], T2)
fs.integrate([nu2], T3)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
return (1 - p_misid) * fs + p_misid * moments.Numerics.reverse_array(fs)
def model_func(params, ns):
s1, t1, nu11, nu12, m1_12, m1_21 = params
sts = moments.LinearSystem_1D.steady_state_1D(np.sum(ns))
fs = moments.Spectrum(sts)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
nu2_func = lambda t: ((1 - s1) * 1.0) * (nu12 / ((1 - s1) * 1.0))**(t / t1)
migs = np.array([[0, m1_12], [m1_21, 0]])
fs.integrate(tf=t1,
Npop=lambda t: [nu11, nu2_func(t)],
m=migs,
dt_fac=0.01)
return fs
def s12IMi(params, ns):
"""
ancestral pop size change, then
split into two populations with exponential growth
Migration is asymmetric and scales with the size of the source population
p_misid: proportion of misidentified ancestral states
Ps: proportion of sites with lower Ne
Fs: factor of Ne reduction (1e-4 - 0.999)
"""
nu0, nu1_0, nu2_0, nu1, nu2, T0, T2, m12, m21 = params
nu1_func = lambda t: nu1_0 * (nu1 / nu1_0)**(t / T2)
nu2_func = lambda t: nu2_0 * (nu2 / nu2_0)**(t / T2)
nu_func = lambda t: [nu1_func(t), nu2_func(t)]
"""
nu1s_func = lambda t: Fs*nu1_0 * (nu1/nu1_0)**(t/T2)
nu2s_func = lambda t: Fs*nu2_0 * (nu2/nu2_0)**(t/T2)
nus_func = lambda t: [nu1s_func(t), nu2s_func(t)]
"""
m21_func = lambda t: m21 * nu2_func(t)
m12_func = lambda t: m12 * nu1_func(t)
migs = lambda t: np.array([[0, m12_func(t)], [m21_func(t), 0]])
migs.i = lambda t: np.array([[0, Fi * m12_func(t)], [Fi * m21_func(t), 0]])
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs.integrate([nu0], T0)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
# fs.integrate([nu1_0, nu2_0], T1, m = np.array([[0, 0], [0, 0]]))
fs.integrate(nu_func, T2, dt_fac=0.01, m=migs)
"""
stsi = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsi = moments.Spectrum(stsi)
fsi.integrate([nu0], T0)
fsi = moments.Manips.split_1D_to_2D(fsi, ns[0], ns[1])
# fsi.integrate([nu1_0, nu2_0], T1, m = np.array([[0, 0], [0, 0]]))
fsi.integrate(nu_func, T2, dt_fac=0.01, m=migs.i)
stss = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fss = moments.Spectrum(stss)
fss.integrate([nu0*Fs], T0)
fss = moments.Manips.split_1D_to_2D(fss, ns[0], ns[1])
# fss.integrate([Fs*nu1_0, Fs*nu2_0], T1, m = np.array([[0, 0], [0, 0]]))
fss.integrate(nus_func, T2, dt_fac=0.01, m=migs)
stsis = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsis = moments.Spectrum(stsis)
fsis.integrate([nu0*Fs], T0)
fsis = moments.Manips.split_1D_to_2D(fsis, ns[0], ns[1])
fsis.integrate([Fs*nu1_0, Fs*nu2_0], T1, m = np.array([[0, 0], [0, 0]]))
fsis.integrate(nus_func, T2, dt_fac=0.01, m=migs.i)
"""
# fs2=Ps*fsi+(1-Pi)*fs
return fs
def model_func(params, ns):
"""
Some model
"""
nuB, nuF, TB, TF = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0])
fs = moments.Spectrum(sts)
fs.integrate(tf=TB, Npop=[nuB])
fs.integrate(tf=TF, Npop=[nuF])
return fs
def sc3ei(params , ns):
# p_misid: proportion of misidentified ancestral states
# P: proportion of sites with lower Ne
# Fs: factor of Ne reduction (1e-5 - 0.999)
nu1,nu1_2,nu2_2,T1,T2,m12,m21,p_misid = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs.integrate([nu1], T1)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1_2, nu2_2], T2, m = np.array([[0, m12], [m21, 0]]))
return (1-p_misid)*fs + p_misid*moments.Numerics.reverse_array(fs)
def SI_2N(params, ns):
"""
nu1= pop size for North Sea
nu2=pop size for Baltic Sea
T1= time of split
hrf= Hill-Robertson factor (i.e. average Ne for regions under linked selection as a proportion of "neutral" Ne)
Q= proportion of the genome under linked selection
"""
nu1,nu2,T1,hrf,Q = params
# calculate the spectrum for neutral portion of the genome
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1, nu2], T1, dt_fac=0.01, m=numpy.array([[0, 0], [0, 0]]))
# calculate the spectrum for lower Ne regions of the genome
stsn2 = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsn2 = moments.Spectrum(stsn2)
fsn2 = moments.Manips.split_1D_to_2D(fsn2, ns[0], ns[1])
fsn2.integrate([nu1*hrf, nu2*hrf], T1, dt_fac=0.01, m=numpy.array([[0, 0], [0, 0]]))
fs2= Q*fsn2+(1-Q)*fs
return fs2
def sc3ei(params, ns):
# p_misid: proportion of misidentified ancestral states
# P: proportion of sites with lower migration
# Fi: factor of migr reduction (1e-5 - 0.999)
nu1, nu1_2, nu2_2, T1, T2, m12, m21, Fi, P = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs.integrate([nu1], T1)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1_2, nu2_2], T2, m=np.array([[0, m12], [m21, 0]]))
stsi = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsi = moments.Spectrum(stsi)
fsi.integrate([nu1], T1)
fsi = moments.Manips.split_1D_to_2D(fsi, ns[0], ns[1])
fsi.integrate([nu1_2, nu2_2],
T2,
m=np.array([[0, m12 * Fi], [m21 * Fi, 0]]))
fs2 = P * fsi + (1 - P) * fs
return fs2
def AM_b_2M(params, ns):
"""
nu1= pop size for North Sea
s=proportion of the North Sea pop which invaded the Baltic (i.e. original bottleneck)
nu2= final size of Baltic Sea pop
T1= time of population split
T2= time of second epoch of migration and population growth
m12= migration rate from North Sea to Baltic
m21= migration rate from Baltic Sea to North Sea
i1= reduction in migration rate for Islands from the Baltic to North Sea
i2= reduction in migration rate for Islands from the North Sea to Baltic
P= proportion of the genome made up of "islands"
O= Proportion of miss-polirized SNPs
"""
nu1, nu2, s, T1, T2, m12, m21, i1, i2, P, O = params
nu2_0 = nu1 * s
nu2_func = lambda t: nu2_0 * (nu2 / nu2_0)**(t / T2)
nu_func = lambda t: [nu1, nu2_func(t)]
# calculate the spectrum for neutral portion of the genome
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsO = moments.Spectrum(sts)
fsO = moments.Manips.split_1D_to_2D(fsO, ns[0], ns[1])
fsO.integrate([nu1, nu1 * s],
T1,
dt_fac=0.01,
m=numpy.array([[0, m12], [m21, 0]]))
fsO.integrate(nu_func, T2, dt_fac=0.01, m=numpy.array([[0, 0], [0, 0]]))
fsM = moments.Numerics.reverse_array(fsO)
# calculate teh spectrum for genomic islands
stsi = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsiO = moments.Spectrum(stsi)
fsiO = moments.Manips.split_1D_to_2D(fsiO, ns[0], ns[1])
fsiO.integrate([nu1, nu1 * s],
T1,
dt_fac=0.01,
m=numpy.array([[0, m12 * i2], [m21 * i1, 0]]))
fsiO.integrate(nu_func, T2, dt_fac=0.01, m=numpy.array([[0, 0], [0, 0]]))
fsiM = moments.Numerics.reverse_array(fsiO)
fs2 = P * (O * fsiO + (1 - O) * fsiM) + (1 - P) * (O * fsO + (1 - O) * fsM)
return fs2
def SI(params, ns):
"""
nu1= pop size for North Sea
nu2=pop size for Baltic Sea
T1= time of split
"""
nu1,nu2,T1 = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1, nu2], T1, dt_fac=0.01,m = numpy.array([[0, 0], [0, 0]]))
return fs
def sc3ei(params, ns):
# p_misid: proportion of misidentified ancestral states
# P: proportion of sites with lower migration
# Fi: factor of migration reduction (1e-5 - 0.999)
nu0, nu1_1, nu2_1, nu1_2, nu2_2, T0, T1, T2, m12_1, m21_1, m12_2, m21_2 = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs.integrate([nu0], T0)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1_1, nu2_1], T1, m=np.array([[0, m12_1], [m21_1, 0]]))
fs.integrate([nu1_2, nu2_2], T2, m=np.array([[0, m12_2], [m21_2, 0]]))
return fs
def IM(params, ns):
s, nPer, nAlb, tSplit, m12, m21 = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
nu1_func = lambda t: s * (nPer / s)**(t / tSplit)
nu2_func = lambda t: (1 - s) * (nAlb / (1 - s))**(t / tSplit)
nu_func = lambda t: [nu1_func(t), nu2_func(t)]
fs.integrate(nu_func,
tSplit,
dt_fac=0.01,
m=np.array([[0, m12], [m21, 0]]))
return fs
def test_folded_slices(self):
ns = (3, 4)
fs1 = moments.Spectrum(numpy.random.rand(*ns))
folded1 = fs1.fold()
self.assert_(fs1[:].folded == False)
self.assert_(folded1[:].folded == True)
self.assert_(fs1[0].folded == False)
self.assert_(folded1[1].folded == True)
self.assert_(fs1[:, 0].folded == False)
self.assert_(folded1[:, 1].folded == True)
def SC_2M(params, ns):
"""
nu1= pop size for North Sea
nu2= pop size for the Baltic Sea
T1= time of population split
T2= time of secondary contact
i1= reduction in migration rate for Islands from the Baltic to North Sea
i2= reduction in migration rate for Islands from the North Sea to Baltic
P= proportion of the genome made up of "islands"
O= Proportion of miss-polirized SNPs
"""
nu1, nu2, T1, T2, m12, m21, i1, i2, P, O = params
# calculate the spectrum for neutral portion of the genome
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsO = moments.Spectrum(sts)
fsO = moments.Manips.split_1D_to_2D(fsO, ns[0], ns[1])
fsO.integrate([nu1, nu2], T1, dt_fac=0.01, m=numpy.array([[0, 0], [0, 0]]))
fsO.integrate([nu1, nu2],
T2,
dt_fac=0.01,
m=numpy.array([[0, m12], [m21, 0]]))
fsM = moments.Numerics.reverse_array(fsO)
# calculate the spectrum from genomic islands
stsi = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsiO = moments.Spectrum(stsi)
fsiO = moments.Manips.split_1D_to_2D(fsiO, ns[0], ns[1])
fsiO.integrate([nu1, nu2],
T1,
dt_fac=0.01,
m=numpy.array([[0, 0], [0, 0]]))
fsiO.integrate([nu1, nu2],
T2,
dt_fac=0.01,
m=numpy.array([[0, m12 * i2], [m21 * i1, 0]]))
fsiM = moments.Numerics.reverse_array(fsiO)
fs2 = P * (O * fsiO + (1 - O) * fsiM) + (1 - P) * (O * fsO + (1 - O) * fsM)
return fs2
def make_plot(sim_fs, integrated_fs, args, initial_seed):
ndemes = len(integrated_fs.shape)
outfile = os.path.basename(args.yaml).replace(".yml", "_residuals.png")
fig = plt.Figure()
if ndemes == 2:
moments_sim_fs = moments.Spectrum(sim_fs.todense())
moments.Plotting.plot_2d_comp_Poisson(
integrated_fs,
moments_sim_fs,
show=False,
fig_num=plt.gcf().number,
)
# simfs = os.path.basename(args.yaml).replace("yml", "mean_sim_fs")
# moments_sim_fs.to_file(simfs)
# simfs = os.path.basename(args.yaml).replace("yml", "integrated_fs")
# integrated_fs.to_file(simfs)
elif ndemes == 1:
moments_sim_fs = moments.Spectrum(sim_fs.data[:-1])
moments.Plotting.plot_1d_comp_Poisson(
integrated_fs[:-1],
moments_sim_fs,
show=False,
fig_num=plt.gcf().number,
)
# simfs = os.path.basename(args.yaml).replace("yml", "mean_sim_fs")
# moments_sim_fs.to_file(simfs)
# simfs = os.path.basename(args.yaml).replace("yml", "integrated_fs")
# integrated_fs.to_file(simfs)
else:
raise NotImplementedError(
f"Plotting not implemented for {ndemes} demes.")
plt.gcf().suptitle(f"No. reps = {args.nreps}, seed = {initial_seed}")
plt.gcf().tight_layout()
plt.savefig(outfile)
return outfile
def sc3ei(params, ns):
# p_misid: proportion of misidentified ancestral states
# P: proportion of sites with lower migration
# Fi: factor of migr reduction (1e-5 - 0.999)
nu1, nu1_2, nu2_2, T1, T2, m12, m21, Fi, Pi, Fs, Ps, p_misid = params
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs.integrate([nu1], T1)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1_2, nu2_2], T2, m=np.array([[0, m12], [m21, 0]]))
stsi = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsi = moments.Spectrum(stsi)
fsi.integrate([nu1], T1)
fsi = moments.Manips.split_1D_to_2D(fsi, ns[0], ns[1])
fsi.integrate([nu1_2, nu2_2],
T2,
m=np.array([[0, m12 * Fi], [m21 * Fi, 0]]))
stsis = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsis = moments.Spectrum(stsis)
fsis.integrate([nu1], T1)
fsis = moments.Manips.split_1D_to_2D(fsis, ns[0], ns[1])
fsis.integrate([nu1_2 * Fs, nu2_2 * Fs],
T2,
m=np.array([[0, m12 * Fi * Fs], [m21 * Fi * Fs, 0]]))
stss = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fss = moments.Spectrum(stss)
fss.integrate([nu1], T1)
fss = moments.Manips.split_1D_to_2D(fss, ns[0], ns[1])
fss.integrate([nu1_2 * Fs, nu2_2 * Fs],
T2,
m=np.array([[0, m12 * Fs], [m21 * Fs, 0]]))
fs2 = Pi * (1 - Ps) * fsi + Ps * (1 - Pi) * fss + Pi * Ps * fsis + (
1 - Pi) * (1 - Ps) * fs
return (1 - p_misid) * fs2 + p_misid * moments.Numerics.reverse_array(fs2)
def SC_b_2N(params, ns):
"""
nu1= pop size for North Sea
s=proportion of the North Sea pop which invaded the Baltic (i.e. original bottleneck)
nu2= final size of Baltic Sea pop
T1= time of population split
T2= time of secondary contact
m12= migration rate from North Sea to Baltic
m21= migration rate from Baltic Sea to North Sea
hrf= Hill-Robertson factor (i.e. average Ne for regions under selection as a proportion of "neutral" Ne)
Q= proportion of the genome under selection
"""
nu1, nu2, s, T1, T2, m12, m21, hrf, Q = params
nu2_0 = nu1 * s
nu2_func = lambda t: nu2_0 * (nu2 / nu2_0)**(t / T2)
nu_func = lambda t: [nu1, nu2_func(t)]
nu2hrf_func = lambda t: nu2_0 * (nu2 / nu2_0)**(t / T2) * hrf
nuhrf_func = lambda t: [nu1 * hrf, nu2hrf_func(t)]
# calculate the spectrum for neutral portion of the genome
sts = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fs = moments.Spectrum(sts)
fs = moments.Manips.split_1D_to_2D(fs, ns[0], ns[1])
fs.integrate([nu1, nu1 * s], T1, m=np.array([[0, 0], [0, 0]]))
fs.integrate(nu_func, T2, dt_fac=0.01, m=np.array([[0, m12], [m21, 0]]))
# calculate the spectrum from low Ne parts of the genome
stsn2 = moments.LinearSystem_1D.steady_state_1D(ns[0] + ns[1])
fsn2 = moments.Spectrum(stsn2)
fsn2 = moments.Manips.split_1D_to_2D(fsn2, ns[0], ns[1])
fsn2.integrate([nu1 * hrf, nu1 * s * hrf],
T1,
m=np.array([[0, 0], [0, 0]]))
fsn2.integrate(nuhrf_func,
T2,
dt_fac=0.01,
m=np.array([[0, m12], [m21, 0]]))
fs2 = Q * fsn2 + (1 - Q) * fs
return fs2
