def test_iqspr_resample1(data):
# not sure if this test can be fully reliable by only fixing the random seed
like_mdl = data['like_mdl']
ngram = data['ngram']
beta = np.linspace(0.1, 1, 2)
np.random.seed(0)
iqspr = IQSPR(estimator=like_mdl, modifier=ngram, r_ESS=0)
soln1 = [['C([*])C([*])(C(=O)OCCSCCC#N)', 'C([*])C([*])(SCCC)',
'O([*])C(=O)OC(C=C1)=CC=C1C(C=C2)=CC=C2CC(C=C3)=CC=C3C(C=C4)=CC=C4([*])'],
['C([*])C([*])(C(=O)OCC(F)(F)C(F)(F)OC(F)(F)OC(F)(F)C(F)(F)OC(F)(F)C(F)(F)F)',
'C([*])C([*])(CC)(C(=O)OCC(F)(F)F)',
'O([*])C(=O)OC(C=C1)=CC=C1C(C=C1)=CC=C1CC(C=C1)=CC=C1C(C=C1)=CC=C1C(C=C1)=CC=C1C(C=C1)=CC=C1C(C=C1)=CC=C1C(=S)']
]
c0 = 0
for s, ll, p, f in iqspr(data['pg'][0][:3], beta, yield_lpf=True):
assert np.abs(np.sum(p) - 1.0) < 1e-5
assert np.sum(f) == 3
assert np.all(np.sort(s) == np.array(soln1[c0]))
c0 += 1
np.random.seed(0)
iqspr = IQSPR(estimator=like_mdl, modifier=ngram, r_ESS=1)
soln2 = [['C([*])C([*])(C(=O)OCCSCCC#N)', 'C([*])C([*])(SCCC)',
'O([*])C(=O)OC(C=C1)=CC=C1C(C=C2)=CC=C2CC(C=C3)=CC=C3C(C=C4)=CC=C4([*])'],
['O([*])C(=O)OC(C=C1)=CC=C1C(C=C1)=CC=C1CC(C=C1)=CC=C1C(C=C1)=CC=C1([*])',
'O([*])C(=O)OC(C=C1)=CC=C1C(C=C1)=CC=C1CC(C=C1)=CC=C1C(C=C1)=CC=C1C(=S)']
]
c0 = 0
for s, ll, p, f in iqspr(data['pg'][0][:3], beta, yield_lpf=True):
assert np.abs(np.sum(p) - 1.0) < 1e-5
assert np.sum(f) == 3
assert np.all(np.sort(s) == np.array(soln2[c0]))
c0 += 1
def run_iqspr(prd_mdls, n_gram, init_samples, beta):
# library for running iQSPR in XenonPy-iQSPR
from xenonpy.inverse.iqspr import IQSPR
# set up likelihood and n-gram models in iQSPR
iqspr = IQSPR(estimator=prd_mdls, modifier=n_gram)
np.random.seed(201903) # fix the random seed
# main loop of iQSPR
samples, loglike, prob, freq = [], [], [], []
for s, ll, p, freq in iqspr(init_samples, beta, yield_lpf=True,
**{'E': (0, 200), 'H**O-LUMO gap': (-np.inf, 3)}):
samples.append(s)
loglike.append(ll)
prob.append(p)
freq.append(freq)
# record all outputs
iqspr_results_reorder = {
"samples": samples,
"loglike": loglike,
"prob": prob,
"freq": freq,
"beta": np.hstack([0, beta]) # include the step of initial samples
}
# save results
with open('iQSPR_results_reorder.obj', 'wb') as f:
pk.dump(iqspr_results_reorder, f)
with open('iQSPR_results_reorder.obj', 'rb') as f:
iqspr_results_reorder = pk.load(f)
return iqspr_results_reorder
def run(self):
from xenonpy.inverse.iqspr import IQSPR
from xenonpy.inverse.iqspr import BayesianRidgeEstimator
self.check_properties()
prd_mdls = BayesianRidgeEstimator(descriptor=self.descriptor_generator,
**self.models)
iqspr = IQSPR(estimator=prd_mdls, modifier=self.ngram_model)
proposed_structures, log_likelihood, probability_score, iqspr_freq = [], [], [], []
for s, ll, p, freq in iqspr(self.seed_structure,
self.beta,
yield_lpf=True,
**self.desired_values):
proposed_structures.append(s)
log_likelihood.append(ll)
probability_score.append(p)
iqspr_freq.append(freq)
# record all outputs
iqspr_results = {
"samples": proposed_structures,
"loglike": log_likelihood,
"prob": probability_score,
"freq": iqspr_freq,
"beta":
np.hstack([0, self.beta]) # include the step of initial samples
}
self.iqspr_results = iqspr_results
return iqspr, iqspr_results
def test_iqspr_2(data):
np.random.seed(0)
like_mdl = data['like_mdl']
ngram = data['ngram']
iqspr = IQSPR(estimator=like_mdl, modifier=ngram)
beta1 = np.linspace(0.05, 1, 10)
beta2 = np.linspace(0.01, 1, 10)
beta = pd.DataFrame({'bandgap': beta1, 'glass_transition_temperature': beta2,
'density': beta1, 'refractive_index': beta2})
for s, ll, p, f in iqspr(data['pg'][0][:5], beta, yield_lpf=True):
assert np.abs(np.sum(p) - 1.0) < 1e-5
assert np.sum(f) == 5
def test_iqspr_1(data):
np.random.seed(0)
ecfp = ECFP(n_jobs=1, input_type='smiles')
bre = BayesianRidgeEstimator(descriptor=ecfp)
ngram = NGram()
iqspr = IQSPR(estimator=bre, modifier=ngram)
X, y = data['pg']
bre.fit(X, y)
ngram.fit(data['pg'][0][0:20], train_order=10)
beta = np.linspace(0.05, 1, 10)
for s, ll, p, f in iqspr(data['pg'][0][:5], beta, yield_lpf=True, bandgap=(0.1, 0.2), density=(0.9, 1.2)):
assert np.abs(np.sum(p) - 1.0) < 1e-5
assert np.sum(f) == 5, print(f)
def test_iqspr_1(data):
np.random.seed(0)
ecfp = data['ecfp']
bre = GaussianLogLikelihood(descriptor=ecfp)
ngram = NGram()
iqspr = IQSPR(estimator=bre, modifier=ngram)
X, y = data['pg']
bre.fit(X, y)
bre.update_targets(reset=True, bandgap=(0.1, 0.2), density=(0.9, 1.2))
ngram.fit(data['pg'][0][0:20], train_order=10)
beta = np.linspace(0.05, 1, 10)
for s, ll, p, f in iqspr(data['pg'][0][:5], beta, yield_lpf=True):
assert np.abs(np.sum(p) - 1.0) < 1e-5
assert np.sum(f) == 5
def data():
# ignore numpy warning
import warnings
print('ignore NumPy RuntimeWarning\n')
warnings.filterwarnings("ignore", message="numpy.dtype size changed")
warnings.filterwarnings("ignore", message="numpy.ndarray size changed")
pwd = Path(__file__).parent
pg_data = pd.read_csv(str(pwd / 'polymer_test_data.csv'))
X = pg_data['smiles']
y = pg_data.drop(['smiles', 'Unnamed: 0'], axis=1)
ecfp = ECFP(n_jobs=1, input_type='smiles', target_col=0)
rdkitfp = RDKitFP(n_jobs=1, input_type='smiles', target_col=0)
bre = GaussianLogLikelihood(descriptor=ecfp)
bre2 = GaussianLogLikelihood(descriptor=rdkitfp)
bre.fit(X, y[['bandgap', 'glass_transition_temperature']])
bre2.fit(X, y[['density', 'refractive_index']])
bre.update_targets(bandgap=(1, 2), glass_transition_temperature=(200, 300))
bre2.update_targets(refractive_index=(2, 3), density=(0.9, 1.2))
class MyLogLikelihood(BaseLogLikelihoodSet):
def __init__(self):
super().__init__()
self.loglike = bre
self.loglike = bre2
like_mdl = MyLogLikelihood()
ngram = NGram()
ngram.fit(X[0:20], train_order=5)
iqspr = IQSPR(estimator=bre, modifier=ngram)
# prepare test data
yield dict(ecfp=ecfp,
rdkitfp=rdkitfp,
bre=bre,
bre2=bre2,
like_mdl=like_mdl,
ngram=ngram,
iqspr=iqspr,
pg=(X, y))
print('test over')
def data():
# ignore numpy warning
import warnings
print('ignore NumPy RuntimeWarning\n')
warnings.filterwarnings("ignore", message="numpy.dtype size changed")
warnings.filterwarnings("ignore", message="numpy.ndarray size changed")
pwd = Path(__file__).parent
pg_data = pd.read_csv(str(pwd / 'polymer_test_data.csv'))
X = pg_data['smiles']
y = pg_data.drop(['smiles', 'Unnamed: 0'], axis=1)
ecfp = ECFP(n_jobs=1, input_type='smiles')
bre = BayesianRidgeEstimator(descriptor=ecfp)
ngram = NGram()
iqspr = IQSPR(estimator=bre, modifier=ngram)
# prepare test data
yield dict(ecfp=ecfp, bre=bre, ngram=ngram, iqspr=iqspr, pg=(X, y))
print('test over')
