rec['resp'] = rec['resp'].rasterize()
rec['stim'] = rec['stim'].rasterize()
rec = mask_high_repetion_stims(rec)
rec = rec.apply_mask(reset_epochs=True)
rec = rec.create_mask(True)
if pupil_regress:
rec = preproc.regress_state2(rec, state_sigs=['pupil'], regress=['pupil'])
rec = rec.and_mask(['PreStimSilence', 'PostStimSilence'], invert=True)
rec = rec.apply_mask(reset_epochs=True)
# Create pupil masks
p_ops = {'state': 'big', 'epoch': ['REFERENCE'], 'collapse': True}
rec_bp = preproc.create_pupil_mask(rec, **p_ops)
p_ops['state'] = 'small'
rec_sp = preproc.create_pupil_mask(rec, **p_ops)
# balanced epochs
epochs = preproc.get_pupil_balanced_epochs(rec, rec_sp, rec_bp)
rec_bp = rec_bp.and_mask(epochs)
rec_sp = rec_sp.and_mask(epochs)
rec_bp = rec_bp.apply_mask(reset_epochs=True)
rec_sp = rec_sp.apply_mask(reset_epochs=True)
# Compute coarse MI for each neuron, over all stimuli (balanced)
Rbig = rec_bp['resp'].extract_epoch('REFERENCE').mean(0).mean(-1)
Rsmall = rec_sp['resp'].extract_epoch('REFERENCE').mean(0).mean(-1)
MI = (Rbig - Rsmall) / (Rbig + Rsmall)
rec['stim'] = rec['stim'].rasterize()
rec = mask_high_repetion_stims(rec)
rec = rec.apply_mask(reset_epochs=True)
rec = preproc.generate_psth(rec)
if balance_stims:
epochs = preproc.get_pupil_balanced_epochs(rec)
rec = rec.and_mask(epochs)
rec = rec.apply_mask(reset_epochs=True)
if evoked_only:
rec = rec.and_mask(['PreStimSilence', 'PostStimSilence'], invert=True)
rec = rec.apply_mask(reset_epochs=True)
pup_ops = {'state': 'big', 'epoch': ['REFERENCE'], 'collapse': True}
rec_bp = preproc.create_pupil_mask(rec.copy(), **pup_ops)
pup_ops['state'] = 'small'
rec_sp = preproc.create_pupil_mask(rec.copy(), **pup_ops)
big_mask = rec_bp['mask']
small_mask = rec_sp['mask']
# Now, fit the model for the range of alpha2 values
nCells = len(rec['resp'].chans)
# intialize fit randomly. On other iterations, previous fit will be new x0
x0 = np.zeros(
nCells * 3
) # because of exp, this effectively initializes gain to 1 and baseline terms to 0
model_output = opt.minimize(glm.pupil_only_objective,
x0, (rec),