def test_fit_numpyro_serial():
df, df_monster, index_cols, firing_rates = generate_fake_spikes(
n_trials=2,
n_neurons=3,
n_mice=4,
dur=2,
)
for y in (set(df.columns) - set(index_cols)):
trace, mcmc = fit_numpyro(
y=df[y].values,
treatment=trans(df['stim']),
condition=trans(df['neuron']),
group=trans(df['mouse']),
progress_bar=True,
model=models.model_hierarchical,
n_draws=100,
num_chains=1,
)
# chart = visualization.plot_posterior_altair(trace,
# df,
# b_name='slope_per_condition',
# x='stim',
# group_name='neuron'
# )
#
# assert ((type(chart) == FacetChart) |
# (type(chart) == Chart) |
# (type(chart) == LayerChart)), print(f'{type(chart)}')
trace.posterior.to_dataframe().pipe(ck.has_no_nans)
def test_fit_lme_w_data_condition():
df, df_monster, index_cols, _ = generate_fake_spikes(n_trials=25)
window = LMERegression(df=df, y='isi', treatment='stim', group='mouse',
condition='neuron_x_mouse')
window.fit(do_make_change='divide', )
window.plot().display()
window.facet(column='neuron_x_mouse', width=300).display()
def test_fake_spikes_explore():
df, df_monster, index_cols, firing_rates = generate_fake_spikes(
n_trials=2,
n_neurons=3,
n_mice=4,
dur=2,
)
charts = fake_spikes_explore(df, df_monster, index_cols)
for chart in charts:
assert ((type(chart) == FacetChart) | (type(chart) == Chart) |
(type(chart) == LayerChart)), print(f'{type(chart)}')
chart.display()
def test_slopes():
df, df_monster, index_cols, firing_rates = generate_fake_spikes(
n_trials=5, n_neurons=3, n_mice=4, dur=7, mouse_response_slope=16)
bw = BayesRegression(df=df,
y='isi',
treatment='stim',
condition='neuron',
group='mouse')
bw.fit(model=models.model_hierarchical,
do_make_change='subtract',
fold_change_index_cols=('stim', 'mouse', 'neuron'))
bw.plot()
def test_plot_data_slope_trials():
df, df_monster, index_cols, firing_rates = generate_fake_spikes(
n_trials=2,
n_neurons=3,
n_mice=4,
dur=2,
)
chart = plot_data_slope_trials(df=df,
x='stim:O',
y='log_firing_rate',
color='neuron:N',
detail='i_trial')
chart.display()
def test_generate_fake_spikes():
df, df_monster, index_cols, firing_rates = generate_fake_spikes()
df = df.set_index(list(index_cols))
assert abs(
firing_rates.sel(Stim=1).isel(Neuron=0, Mouse=0) -
firing_rates.sel(Stim=0).isel(Neuron=0, Mouse=0)) < 1
assert (firing_rates.sel(Stim=1).isel(Neuron=0, Mouse=-1) -
firing_rates.sel(Stim=0).isel(Neuron=0, Mouse=-1)) > 1
assert abs(
firing_rates.sel(Stim=1).isel(Neuron=-1, Mouse=-1) -
firing_rates.sel(Stim=0).isel(Neuron=-1, Mouse=-1)) < 1
df.pipe(ck.has_no_nans)
df_monster.pipe(ck.has_no_nans)
def test_fit_lme_w_condition():
from numpy.linalg import LinAlgError
df, df_monster, index_cols, firing_rates = generate_fake_spikes(n_trials=20,
n_neurons=7,
n_mice=6,
dur=7,
mouse_response_slope=12,
overall_stim_response_strength=45)
try:
reg = LMERegression(df=df, y='isi', treatment='stim', condition='neuron_x_mouse', group='mouse', )
assert reg.fit().data_and_posterior is not None
reg.plot(x=reg.window.condition[0]).display()
reg.plot(x='neuron_x_mouse').display()
reg.facet(column='neuron_x_mouse', width=300).display()
assert len(reg.charts) > 0
except LinAlgError as e:
print(e)
def test_compare_models():
df, df_monster, index_cols, firing_rates = generate_fake_spikes(
n_trials=10,
n_neurons=10,
n_mice=4,
dur=2,
)
compare_models(
df=df,
models={
'no_neuron': models.model_hierarchical,
'no_neuron_or_treatment': models.model_hierarchical,
'no-treatment': models.model_hierarchical,
'treatment': models.model_hierarchical,
'student': models.model_hierarchical,
'lognogmal': models.model_hierarchical,
},
extra_model_args=[{
'treatment': 'stim',
'condition': None
}, {
'treatment': None,
'condition': None
}, {
'treatment': None,
'condition': 'neuron'
}, {
'treatment': 'stim',
'condition': 'neuron'
}, {
'treatment': 'stim',
'condition': 'neuron',
'dist_y': 'student'
}, {
'treatment': 'stim',
'condition': 'neuron',
'dist_y': 'lognormal'
}],
y='isi',
group='mouse',
parallel=True,
plotose=True,
# num_chains=1,
num_warmup=100,
n_draws=1000,
)
def test_fit_numpyro_gamma():
df, df_monster, index_cols, firing_rates = generate_fake_spikes(
n_trials=2,
n_neurons=3,
n_mice=4,
dur=2,
)
trace, mcmc = fit_numpyro(
y=df['isi'].values,
treatment=trans(df['stim']),
condition=trans(df['neuron']),
group=trans(df['mouse']),
progress_bar=True,
model=models.model_hierarchical,
dist_y='gamma',
n_draws=100,
num_chains=1,
)
def test_fit_numpyro_parallel():
df, df_monster, index_cols, firing_rates = generate_fake_spikes()
# df = df.set_index(list(index_cols))
meas = (set(df.columns) - set(index_cols))
traces = Parallel(
n_jobs=-1,
verbose=2,
# backend='multiprocessing'
)(delayed(fit_numpyro)(y=y,
treatment=(df['stim']).astype(int).values,
condition=trans(df['neuron']),
group=trans(df['mouse']),
progress_bar=False,
model=models.model_hierarchical,
n_draws=10) for y in [df[y].values for y in meas])
assert type(traces[0][0].posterior) == xr.Dataset
[
trace[0].posterior.to_dataframe().pipe(ck.has_no_nans)
for trace in traces
]
def test_estimate_posteriors_slope_uneven_n_data_per_condition():
# Trying to reproduce Uneven number of entries in conditions! Try setting do_take_mean=True
df, df_monster, index_cols, firing_rates = generate_fake_spikes(
n_trials=10,
n_neurons=3,
n_mice=4,
dur=2,
)
df = df.drop(df[(df['i_trial'] == 0) &
# (df['neuron_x_mouse'] == '0m0bayes') &
(df['stim'] == 0)].index)
window = BayesRegression(df=df,
y='isi',
treatment='stim',
condition='neuron',
group='mouse',
add_data=True)
window.fit(models.model_hierarchical, do_make_change='divide')
chart = window.plot(x='neuron', column='neuron', row='mouse')
chart.display()
chart = window.plot(x='neuron', column='neuron', row='mouse')
chart.display()
from bayes_window.visualization import plot_posterior
from bayes_window.utils import load_radon
from pytest import mark
import os
os.environ['bayes_window_test_mode'] = 'True'
df_radon = load_radon()
dfl, _, _, _ = generate_fake_lfp(n_trials=5)
df, df_monster, index_cols, firing_rates = generate_fake_spikes(
n_trials=2,
n_neurons=3,
n_mice=4,
dur=4,
)
# @mark.parametrize('add_data', [True, False])
# @mark.parametrize('add_box', [True, False])
@mark.parametrize('add_condition_slope', [False, True])
@mark.parametrize('add_group_slope', [False, True])
# @mark.parametrize('do_mean_over_trials', [True, False])
@mark.parametrize('do_make_change', ['subtract', 'divide', False])
def test_radon(
# add_data, add_box,
add_condition_slope,
add_group_slope,
# do_mean_over_trials,