def test_its_bhmm(self):
BHMM = msm.bayesian_hidden_markov_model([self.double_well_data.dtraj_T100K_dt10_n6good], 2, 10)
self.ck = BHMM.cktest(mlags=[0,1,10])
estref = np.array([[[ 1., 0. ],
[ 0., 1. ]],
[[ 0.98497185, 0.01502815],
[ 0.01459256, 0.98540744]],
[[ 0.88213404, 0.11786596],
[ 0.11877379, 0.88122621]]])
predref = np.array([[[ 1., 0. ],
[ 0., 1. ]],
[[ 0.98497185, 0.01502815],
[ 0.01459256, 0.98540744]],
[[ 0.86824695, 0.13175305],
[ 0.1279342, 0.8720658 ]]])
predLref = np.array([[[ 1. , 0. ],
[ 0. , 1. ]],
[[ 0.98282734, 0.01284444],
[ 0.0123793 , 0.98296742]],
[[ 0.8514399 , 0.11369687],
[ 0.10984971, 0.85255827]]])
predRref = np.array([[[ 1. , 0. ],
[ 0. , 1. ]],
[[ 0.98715575, 0.01722138],
[ 0.0178059 , 0.98762081]],
[[ 0.8865478 , 0.14905352],
[ 0.14860461, 0.89064809]]])
# rough agreement
assert np.allclose(self.ck.estimates, estref, rtol=0.1, atol=10.0)
assert self.ck.estimates_conf[0] is None
assert self.ck.estimates_conf[1] is None
assert np.allclose(self.ck.predictions, predref, rtol=0.1, atol=10.0)
assert np.allclose(self.ck.predictions[0], predLref, rtol=0.1, atol=10.0)
assert np.allclose(self.ck.predictions[1], predRref, rtol=0.1, atol=10.0)
def test_submodel_simple(self):
# sanity check for submodel;
# call should not alter self
from copy import deepcopy
# dtrj = np.random.randint(0, 2, size=100)
# dtrj[np.random.randint(0, dtrj.shape[0], 3)] = 2
# hard-coded due to stochastic failures
dtrj = [
1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 1, 0, 0, 0, 0, 1, 0, 1, 1, 0, 1,
1, 0, 0, 0, 0, 2, 1, 0, 1, 1, 0, 1, 0, 1, 1, 1, 0, 1, 0, 1, 1, 0,
1, 1, 0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 0, 1, 1, 0, 1, 1, 0, 0,
1, 1, 2, 0, 0, 1, 1, 2, 0, 1, 1, 1, 0, 1, 1, 0, 1, 1, 0, 1, 1, 0,
0, 1, 1, 0, 0, 0, 1, 1, 0, 1, 1, 0
]
h = bayesian_hidden_markov_model(dtrj, 3, 2)
h_original = deepcopy(h)
hs = h.submodel_largest(mincount_connectivity=5)
self.assertTrue(h == h_original)
self.assertEqual(hs.timescales().shape[0], 1)
self.assertEqual(hs.pi.shape[0], 2)
self.assertEqual(hs.transition_matrix.shape, (2, 2))
def test_separate_states(self):
dtrajs = [np.array([0, 1, 1, 1, 1, 1, 0, 1, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 1]),
np.array([2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 2]),]
hmm_bayes = bayesian_hidden_markov_model(dtrajs, 3, lag=1, separate=[0], nsamples=100)
# we expect zeros in all samples at the following indexes:
pobs_zeros = [[0, 1, 2, 2, 2], [0, 0, 1, 2, 3]]
for s in hmm_bayes.samples:
assert np.allclose(s.observation_probabilities[pobs_zeros], 0)
for strajs in hmm_bayes.sampled_trajs:
assert strajs[0][0] == 2
assert strajs[0][6] == 2
def setUpClass(cls):
# load observations
import pyemma.datasets
obs = pyemma.datasets.load_2well_discrete().dtraj_T100K_dt10
# hidden states
cls.nstates = 2
# samples
cls.nsamples = 100
cls.lag = 10
cls.bhmm = bayesian_hidden_markov_model([obs], cls.nstates, cls.lag, reversible=True, nsamples=cls.nsamples)
def setUpClass(cls):
# load observations
import pyemma.datasets
obs = pyemma.datasets.load_2well_discrete().dtraj_T100K_dt10
# don't print
import bhmm
bhmm.config.verbose = False
# hidden states
cls.nstates = 2
# samples
cls.nsamples = 100
cls.lag = 10
cls.sampled_hmm_lag10 = bayesian_hidden_markov_model(
[obs],
cls.nstates,
cls.lag,
reversible=True,
nsamples=cls.nsamples)
mplt.plot_cktest(MSM.cktest(3))
plt.savefig('cktest_msm.png')
plt.clf()
plt.figure(figsize=(8, 5))
mplt.plot_free_energy(np.hstack(Y1),
np.hstack(Y2),
weights=np.hstack(MSM.trajectory_weights()))
plt.xlabel('tic 1')
plt.ylabel('tic 2')
plt.savefig('reweighted_tic1_tic2.png')
print('running hmm')
HMM = msm.bayesian_hidden_markov_model(clkmeans.dtrajs, n_macrostates, lag=100)
hmm_dist = HMM.metastable_distributions
hmm_sets = HMM.metastable_sets
plt.clf()
plt.figure(figsize=(8, 5))
mplt.plot_free_energy(np.hstack(Y1),
np.hstack(Y2),
weights=np.hstack(MSM.trajectory_weights()))
colors = [
'red', 'cyan', 'blue', 'purple', 'magenta', 'brown', 'black', 'white',
'orange', 'pink', 'yellowgreen'
]
for i in range(n_macrostates):
plt.scatter(clkmeans.clustercenters[hmm_sets[i], 0],
clkmeans.clustercenters[hmm_sets[i], 1],
import pyemma.coordinates as coor
import pyemma.msm as msm
import pyemma.plots as mplt
Y = np.load('tica_projection.npy')
Y1 = [y[:, 0] for y in Y]
Y2 = [y[:, 1] for y in Y]
clkmeans_clustercenters = np.load('clkmeans_clustercenters.npy')
clkmeans_dtrajs = np.load('clkmeans_dtrajs.npy')
clkmeans_dtrajs = clkmeans_dtrajs.tolist()
HMM = msm.bayesian_hidden_markov_model(clkmeans_dtrajs, 2, lag=100)
state_1_colors = HMM.metastable_distributions[0] / max(
HMM.metastable_distributions[0])
state_2_colors = HMM.metastable_distributions[1] / max(
HMM.metastable_distributions[1])
path_to_trajs = '../MSM-analysis/DDR1_WT_pro_plus_OG_trajectories/*.h5'
from glob import glob
filenames = sorted(glob(path_to_trajs))
print('These are our trajectories that flip.')
print(filenames[16], filenames[42], filenames[145], filenames[149])
import seaborn as sns
system = 'WT-pro'
Y = np.load('%s/tica_projection.npy' % file_paths[system])
Y1 = [y[:, 0] for y in Y]
Y2 = [y[:, 1] for y in Y]
clkmeans_clustercenters = np.load('%s/clkmeans_clustercenters.npy' %
file_paths[system])
clkmeans_dtrajs = np.load('%s/clkmeans_dtrajs.npy' % file_paths[system])
clkmeans_dtrajs = clkmeans_dtrajs.tolist()
HMM = msm.bayesian_hidden_markov_model(clkmeans_dtrajs,
nstates=2,
lag=100,
nsamples=1000,
mincount_connectivity=100)
hmm_sets = HMM.metastable_sets
state0_samples = [
clkmeans_clustercenters[hmm_sets[0], 0],
clkmeans_clustercenters[hmm_sets[0], 1]
]
state1_samples = [
clkmeans_clustercenters[hmm_sets[1], 0],
clkmeans_clustercenters[hmm_sets[1], 1]
]
X_HMM = np.concatenate((state0_samples, state1_samples), axis=1)
