def test_maskarrayp_changeQ_correctentriesmasked(self):
tree = ivy.tree.read(u'(((A:1,B:1)C:1,D:2)E:1,F:3)root;')
chars = [1, 0, 0, 0]
Q1 = np.array([[-0.10, 0.10], [0.05, -0.05]])
Q2 = np.array([[-1.5, 1.5], [1., -1.]])
Q3 = np.array([[-0.15, 0.15], [0.1, -0.1]])
Qs1 = np.array([Q2, Q1])
Qs2 = np.array([Q2, Q3])
switchpoint = [(tree["C"], 0.75)]
ar = mk_mr.create_mkmr_mb_ar(tree, chars, 2)
l1 = mk_mr.mk_mr_midbranch(tree,
chars,
Qs1,
switchpoint,
ar=ar,
debug=False)
l2 = mk_mr.mk_mr_midbranch(tree,
chars,
Qs2,
switchpoint,
ar=ar,
debug=False)
ar2 = mk_mr.create_mkmr_mb_ar(tree, chars, 2)
l2True = mk_mr.mk_mr_midbranch(tree, chars, Qs2, switchpoint, ar=ar2)
self.assertEqual(l2, l2True)
def foo(tree=tree,
chars=chars,
Qs1=Qs1,
switchpoint_1=switchpoint_1,
ar=ar):
mk_mr.mk_mr_midbranch(tree,
chars,
Qs1,
switchpoint_1,
ar=ar,
debug=False)
def test_makemklnlfunc_twoswitch_makesfunc(self):
tree = ivy.tree.read(u'(((A:1,B:1)C:1,D:2)E:1,F:3)root;')
data = {"A": 1, "B": 0, "D": 0, "F": 0}
Q1 = np.array([[-0.10, 0.10], [0.05, -0.05]])
Q2 = np.array([[-1.5, 1.5], [1., -1.]])
Q3 = np.array([[-.01, .01], [.015, -.015]])
Qs = np.array([Q2, Q3, Q1])
switchpoint = [(tree["C"], 0.75), (tree["E"], 0.25)]
qidx = np.array([[0, 0, 1, 1], [0, 1, 0, 0], [1, 0, 1, 3],
[1, 1, 0, 2], [2, 0, 1, 4], [2, 1, 0, 5]],
dtype=np.intp)
switches = np.array([s[0].ni for s in switchpoint], dtype=np.intp)
lengths = np.array([s[1] for s in switchpoint], dtype=np.double)
f = cyexpokit.make_mklnl_func(tree, data, 2, 3, qidx)
params = np.array([0.05, 0.1, 1.0, 1.5, .01, .015])
cylik = f(params, switches, lengths)
truelik = mk_mr.mk_mr_midbranch(tree, data, Qs, switchpoint)
self.assertTrue(np.isclose(cylik, truelik))
# If we re-use the same likelihood function, do we get the correct
# likelihood?
switches_2 = np.array([2, 6], dtype=np.int)
lens_2 = np.array([.333, .5])
f_2 = cyexpokit.make_mklnl_func(tree, data, 2, 3, qidx)
cylik1 = f(params, switches_2, lens_2)
cylik2 = f_2(params, switches_2, lens_2)
self.assertTrue(np.isclose(cylik1, cylik2))
def test_makemklnlfunc_multi_makesfunc(self):
tree = ivy.tree.read(u'(((A:1,B:1)C:1,D:2)E:1,F:3)root;')
data = {"A": 1, "B": 0, "D": 0, "F": 0}
Q1 = np.array([[-0.10, 0.10], [0.05, -0.05]])
Q2 = np.array([[-1.5, 1.5], [1., -1.]])
Qs = np.array([Q2, Q1])
switchpoint = (tree["C"], 0.75)
qidx = np.array(
[[0, 0, 1, 1], [0, 1, 0, 0], [1, 0, 1, 3], [1, 1, 0, 2]],
dtype=np.intp)
switches = np.array([switchpoint[0].ni], dtype=np.intp)
lengths = np.array([switchpoint[1]], dtype=np.double)
switches = np.array([tree[switchpoint[0].id].ni], dtype=np.intp)
lengths = np.array([switchpoint[1]], dtype=np.double)
f = cyexpokit.make_mklnl_func(tree, data, 2, 2, qidx)
params = np.array([0.05, 0.1, 1.0, 1.5])
cylik = f(params, switches, lengths)
truelik = mk_mr.mk_mr_midbranch(tree, data, Qs, [switchpoint])
trueinds = np.array([0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0])
self.assertTrue((trueinds == f.qi).all())
self.assertTrue(np.isclose(cylik, truelik))
def test_mkmr_largetree_correctlikelihood(self):
tree = ivy.tree.read(u"support/hrm_600tips.newick")
chars = [
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 0, 1,
1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 0, 1, 0, 1, 0, 1,
0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0,
0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0,
0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 1, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0
]
Q0 = np.array([[-0.061457569587587413, 0.061457569587587413],
[0.061457569587587413, -0.061457569587587413]])
Q1 = np.array([[-1.4976115055655292, 1.4976115055655292],
[1.4976115055655292, -1.4976115055655292]])
Q2 = np.array([[-0.0014644343303779842, 0.0014644343303779842],
[0.0014644343303779842, -0.0014644343303779842]])
Qs = np.array([Q0, Q1, Q2])
switchpoint0 = (tree[579], 1.2)
switchpoint1 = (tree[329], 3.0)
true_L = -89.213330113632566
ar = mk_mr.create_mkmr_mb_ar(tree, chars, 3)
calculated_l = mk_mr.mk_mr_midbranch(tree,
chars,
Qs, [switchpoint0, switchpoint1],
ar=ar,
debug=True)
self.assertTrue(np.isclose(calculated_l, true_L))
def test_likelihoodfunctionmods_correctlikelihood(self):
tree = ivy.tree.read(u'(((A:1,B:1)C:1,D:2)E:1,F:3)root;')
chars = [1, 0, 0, 0]
Q1 = np.array([[-0.10, 0.10], [0.05, -0.05]])
Q2 = np.array([[-1.5, 1.5], [1., -1.]])
Qs = np.array([Q2, Q1])
lf = mk_mr.lf_mk_mr_midbranch_mods(tree,
chars,
mods=[(4, 3), (2, 1)],
findmin=False)
switchpoint = (tree["C"], 0.75)
Qparams = np.array([0.05, 0.1, 1.0, 1.5])
lf_likelihood = lf(Qparams, [switchpoint])
trueLikelihood = mk_mr.mk_mr_midbranch(tree, chars, Qs, [switchpoint])
self.assertTrue(np.isclose(lf_likelihood, trueLikelihood))
def test_makemklnlfunc_makesfunc(self):
tree = ivy.tree.read(
u"((t2:0.3778728602,(t3:0.03239763423,t4:0.03239763423):0.345475226):0.9831289164,t1:1.361001777);"
)
chars = [0, 0, 0, 1]
labels = [lf.label for lf in tree.leaves()]
data = dict(zip(labels, chars))
qidx = np.array([[0, 0, 1, 1], [0, 1, 0, 1]], dtype=np.intp)
f = cyexpokit.make_mklnl_func(tree, data, 2, 1, qidx)
params = np.array([0.1, 0.1])
cylik = f(params, np.array([], dtype=int), np.array([]))
Qs = np.array([[[-0.1, 0.1], [0.1, -0.1]]])
truelik = mk_mr.mk_mr_midbranch(tree, chars, Qs, [])
self.assertTrue(np.isclose(cylik, truelik))
def test_mkmr_middleofbranch_matchesbyhand(self):
tree = ivy.tree.read(u'(((A:1,B:1)C:1,D:2)E:1,F:3)root;')
chars = {"A": 1, "B": 0, "D": 0, "F": 0}
Q1 = np.array([[-0.10, 0.10], [0.05, -0.05]])
Q2 = np.array([[-1.5, 1.5], [1., -1.]])
Qs = np.array([Q2, Q1])
switchpoint = (tree["C"], 0.75)
PA = [[0.449251, 0.550749], [0.367166, 0.632834]] #a
PB = [[0.449251, 0.550749], [0.367166, 0.632834]] #b
PCA = [
[0.49201298, 0.50798702], # Closer to tip, fast regime
[0.33865801, 0.66134199]
]
PCB = [
[0.97546295, 0.02453705], # Closer to root, slow regime
[0.01226853, 0.98773147]
]
PD = [[0.82721215, 0.17278785], [0.08639393, 0.91360607]]
PE = [[0.90713865, 0.09286135], [0.04643067, 0.95356933]]
PF = [[0.75841877, 0.24158123], [0.12079062, 0.87920938]]
L0A = 0
L1A = 1
L0B = 1
L1B = 0
L0D = 1
L1D = 0
L0F = 1
L1F = 0
L0CA = (PA[0][0] * L0A + PA[0][1] * L1A) * (PB[0][0] * L0B +
PB[0][1] * L1B)
L1CA = (PA[1][0] * L0A + PA[1][1] * L1A) * (PB[1][0] * L0B +
PB[1][1] * L1B)
L0CB = PCA[0][0] * L0CA + PCA[0][1] * L1CA
L1CB = PCA[1][0] * L0CA + PCA[1][1] * L1CA
L0E = (PCB[0][0] * L0CB + PCB[0][1] * L1CB) * (PD[0][0] * L0D +
PD[0][1] * L1D)
L1E = (PCB[1][0] * L0CB + PCB[1][1] * L1CB) * (PD[1][0] * L0D +
PD[1][1] * L1D)
L0r = (PE[0][0] * L0E + PE[0][1] * L1E) * (PF[0][0] * L0F +
PF[0][1] * L1F)
L1r = (PE[1][0] * L0E + PE[1][1] * L1E) * (PF[1][0] * L0F +
PF[1][1] * L1F)
predictedLikelihood = math.log(L0r * 0.5 + L1r * 0.5)
calculatedLikelihood = mk_mr.mk_mr_midbranch(tree,
chars,
Qs, [switchpoint],
debug=False)
self.assertTrue(np.isclose(predictedLikelihood, calculatedLikelihood))
def test_mkmr_middleofbranchtwoswitch_matchesbyhand(self):
tree = ivy.tree.read(u'(((A:1,B:1)C:1,D:2)E:1,F:3)root;')
chars = [1, 0, 0, 0]
Q1 = np.array([[-0.10, 0.10], [0.05, -0.05]])
Q2 = np.array([[-1.5, 1.5], [1., -1.]])
Q3 = np.array([[-.01, .01], [.015, -.015]])
Qs = np.array([Q2, Q3, Q1])
switchpoint = [(tree["C"], 0.75), (tree["E"], 0.25)]
PA = [[0.449251, 0.550749], [0.367166, 0.632834]] #a
PB = [[0.449251, 0.550749], [0.367166, 0.632834]] #b
PCA = [
[0.49201298, 0.50798702], # Closer to tip, fast regime
[0.33865801, 0.66134199]
]
PCB = [
[0.99749845, 0.0024922], # Closer to root, slow regime
[0.00373829, 0.99625235]
]
PD = [[0.98049177, 0.01950823], [0.02926235, 0.97073765]]
PEA = [[0.99749845, 0.0024922], [0.00373829, 0.99625235]]
PEB = [[0.9290649, 0.0709351], [0.03546755, 0.96453245]]
PF = [[0.75841877, 0.24158123], [0.12079062, 0.87920938]]
L0A = 0
L1A = 1
L0B = 1
L1B = 0
L0D = 1
L1D = 0
L0F = 1
L1F = 0
L0CA = (PA[0][0] * L0A + PA[0][1] * L1A) * (PB[0][0] * L0B +
PB[0][1] * L1B)
L1CA = (PA[1][0] * L0A + PA[1][1] * L1A) * (PB[1][0] * L0B +
PB[1][1] * L1B)
L0CB = PCA[0][0] * L0CA + PCA[0][1] * L1CA
L1CB = PCA[1][0] * L0CA + PCA[1][1] * L1CA
L0EA = (PCB[0][0] * L0CB + PCB[0][1] * L1CB) * (PD[0][0] * L0D +
PD[0][1] * L1D)
L1EA = (PCB[1][0] * L0CB + PCB[1][1] * L1CB) * (PD[1][0] * L0D +
PD[1][1] * L1D)
L0EB = PEA[0][0] * L0EA + PEA[0][1] * L1EA
L1EB = PEA[1][0] * L0EA + PEA[1][1] * L1EA
L0r = (PEB[0][0] * L0EB + PEB[0][1] * L1EB) * (PF[0][0] * L0F +
PF[0][1] * L1F)
L1r = (PEB[1][0] * L0EB + PEB[1][1] * L1EB) * (PF[1][0] * L0F +
PF[1][1] * L1F)
predictedLikelihood = math.log(L0r * 0.5 + L1r * 0.5)
calculatedLikelihood = mk_mr.mk_mr_midbranch(tree, chars, Qs,
switchpoint)
self.assertTrue(np.isclose(predictedLikelihood, calculatedLikelihood))
qp3 = mod_r3.trace("Qparam_2")[150]
eQ1 = np.array([[-qp1,qp1],
[qp1,-qp1]])
eQ2 = np.array([[-qp2,qp2],
[qp2,-qp2]])
eQ3 = np.array([[-qp3,qp3],
[qp3,-qp3]])
est_Qs = np.array([eQ2,eQ3,eQ1])
s1 = mod_r3.trace("switch_0")[150]
s2 = mod_r3.trace("switch_1")[150]
mk_mr.mk_mr_midbranch(tree, chars, est_Qs, [s1,s2])
l = mod_r3.trace("likelihood")[150]
x0 = mod_r3.trace(str("switch_0")[:])
fig = treefig(tree)
fig.add_layer(ivy.vis.layers.add_tree_heatmap, x0, store="switch0")
x1 = mod_r3.trace(str("switch_1")[:])
fig.add_layer(ivy.vis.layers.add_tree_heatmap, x0, store="switch1",color="blue")
s0 = [s[0].ni for s in x0]
s1 = [s[0].ni for s in x1]
plt.plot(s0)