def getTwoFrameKSStatByRegion(region, all_bin_frame, strong_bin_frame):
best_stim = rc.getBestStimFromRegion(strong_bin_frame, region)
all_bin_stim_frame = all_bin_frame[(all_bin_frame.stim_id == best_stim)
& (all_bin_frame.region == region)]
strong_bin_stim_frame = strong_bin_frame[
(strong_bin_frame.stim_id == best_stim)
& (strong_bin_frame.region == region)]
corr_stat, corr_p_val = ks_2samp(all_bin_stim_frame.corr_coef,
strong_bin_stim_frame.corr_coef)
info_stat, info_p_val = ks_2samp(all_bin_stim_frame.mutual_info_qe,
strong_bin_stim_frame.mutual_info_qe)
return region, corr_stat, corr_p_val, info_stat, info_p_val
def getRegionToSampleDict(correlation_frame, bin_width, measure):
region_to_sample = {}
for region in rc.regions:
best_stim = rc.getBestStimFromRegion(correlation_frame, region)
region_bin_frame = correlation_frame[
(correlation_frame.region == region)
& (correlation_frame.bin_width == bin_width) &
(correlation_frame.stim_id == best_stim)]
print(dt.datetime.now().isoformat() + ' INFO: ' + 'Mean ' + measure +
' = ' + str(region_bin_frame[measure].mean()))
print(dt.datetime.now().isoformat() + ' INFO: ' + 'Median ' + measure +
' = ' + str(region_bin_frame[measure].median()))
region_to_sample[region] = region_bin_frame.loc[:, measure]
return region_to_sample
def main():
print(dt.datetime.now().isoformat() + ' INFO: ' +
'Starting main function...')
print(dt.datetime.now().isoformat() + ' INFO: ' +
'Loading single and pairwise measurements...')
firing_frame, pairwise_frame = getDataFrames(args.single_file,
args.paired_file,
args.bin_width)
if args.plot_type == 'firing':
for region in rc.regions:
best_stim = rc.getBestStimFromRegion(firing_frame, region)
firing_region_frame = firing_frame[
(firing_frame.region == region)
& (firing_frame.stim_id == best_stim)]
pairwise_region_frame = pairwise_frame[
(pairwise_frame.region == region)
& (pairwise_frame.stim_id == best_stim)]
working_frame = getWorkingFrame(firing_region_frame,
pairwise_region_frame)
working_frame.loc[:, 'geometric_mean'] = np.sqrt(
working_frame.first_firing_rate *
working_frame.second_firing_rate)
plotMeasuresVsGeomMean(working_frame, region, best_stim,
args.prefix)
elif args.plot_type == 'pairwise':
max_mi = pairwise_frame.mutual_info_qe.max()
for region in rc.regions:
best_stim = rc.getBestStimFromRegion(pairwise_frame, region)
pairwise_region_frame = pairwise_frame[
(pairwise_frame.region == region)
& (pairwise_frame.stim_id == best_stim)]
plotInfoVsCorr(pairwise_region_frame, region, max_mi, best_stim,
args.prefix)
else:
sys.exit(dt.datetime.now().isoformat() + 'ERROR: ' +
'Unrecognised measure type!')
print(dt.datetime.now().isoformat() + ' INFO: ' + 'Done.')
def main():
print(dt.datetime.now().isoformat() + ' INFO: ' +
'Starting main function...')
print(dt.datetime.now().isoformat() + ' INFO: ' +
'Loading all_regions_stims_pairs_widths.csv...')
correlation_frame = pd.read_csv(os.path.join(csv_dir, args.filename))
if args.correlation_type == 'spike_count':
for region in rc.regions:
best_stim = rc.getBestStimFromRegion(correlation_frame, region)
plotAbsCorrCoefByBinWidthForRegionStim(correlation_frame, region,
best_stim,
args.image_file_prefix,
args.x_axis_scale)
elif args.correlation_type == 'signal':
for region in rc.regions:
plotAbsSignalCorrByBinWidth(correlation_frame, region,
args.image_file_prefix,
args.x_axis_scale)
elif args.correlation_type == 'bifurcation':
for region in rc.regions:
best_stim = rc.getBestStimFromRegion(correlation_frame, region)
plotBifurcatedCorrelationsByBinWidth(correlation_frame, region,
best_stim,
args.image_file_prefix,
args.x_axis_scale)
elif args.correlation_type == 'mutual_info':
plotMutualInfoByBinWidth(correlation_frame, args.image_file_prefix,
args.x_axis_scale)
elif args.correlation_type == 'firing_rate':
plotFiringRateByBinWidth(correlation_frame, args.image_file_prefix)
elif args.correlation_type == 'symm_unc':
for region in rc.regions:
plotSymmUncByBinWidth(correlation_frame, region,
args.image_file_prefix, args.x_axis_scale)
else:
sys.exit(dt.datetime.now().isoformat() + ' ERROR: ' +
'Unrecognised correlation type.')
def plotSymmUncByBinWidth(correlation_frame, region, prefix, x_axis_scale):
best_stim = rc.getBestStimFromRegion(correlation_frame, region)
region_stim_frame = correlation_frame[
(correlation_frame.region == region)
& (correlation_frame.stim_id == best_stim)]
plotColumnByBinWidth(region_stim_frame,
'symm_unc_qe',
region,
x_axis_scale,
r'$U(X;Y)$ (a.u.)',
y_lim=[0, 1])
filename = prefix + 'symm_unc_by_bin_width_' + region + '_' + str(
best_stim) + '.png'
print(dt.datetime.now().isoformat() + ' INFO: ' + 'Saving ' + filename +
'...')
plt.savefig(
os.path.join(image_dir, 'symm_unc_vs_bin_width', x_axis_scale,
filename))
plt.close()
def plotFiringRateByBinWidth(firing_frame, prefix):
y_lim = [0, firing_frame.firing_rate_mean.max()]
for region in rc.regions:
best_stim = rc.getBestStimFromRegion(firing_frame, region)
region_stim_frame = firing_frame[(firing_frame.region == region)
& (firing_frame.stim_id == best_stim)]
plotColumnByBinWidth(region_stim_frame,
'firing_rate_mean',
region,
'linear',
'Firing Rate (Hz)',
y_lim=y_lim)
filename = prefix + 'firing_rate_by_bin_width_' + region + '_' + str(
best_stim) + '.png'
print(dt.datetime.now().isoformat() + ' INFO: ' + 'Saving ' +
filename + '...')
plt.savefig(
os.path.join(image_dir, 'firing_rate_vs_bin_width', filename))
plt.close()
def getBestStimsLims(corr_bin_frame,
strong_corr_bin_frame,
measure,
hist_range=None):
hist_range = [0, corr_bin_frame[measure].max()
] if hist_range == None else hist_range
best_stims = np.zeros(len(rc.regions), dtype=int)
top_count = 0
max_bin = 0
for i, region in enumerate(rc.regions):
best_stims[i] = rc.getBestStimFromRegion(strong_corr_bin_frame, region)
corr_region_bin_stim = corr_bin_frame[
(corr_bin_frame.stim_id == best_stims[i])
& (corr_bin_frame.region == region)]
counts, bins = np.histogram(corr_region_bin_stim[measure],
bins=25,
range=hist_range)
top_count = np.max([top_count, counts.max()])
max_bin = np.max([max_bin, bins[-1]])
return best_stims, [0, 10 * np.ceil(top_count / 10).astype(int)], max_bin
def plotMutualInfoByBinWidth(correlation_frame, prefix, x_axis_scale):
y_lim = [0, correlation_frame.mutual_info_qe.max()]
for region in rc.regions:
best_stim = rc.getBestStimFromRegion(correlation_frame, region)
region_stim_frame = correlation_frame[
(correlation_frame.region == region)
& (correlation_frame.stim_id == best_stim)]
plotColumnByBinWidth(region_stim_frame,
'mutual_info_qe',
region,
x_axis_scale,
r'$I(X;Y)$ (bits)',
y_lim=y_lim)
filename = prefix + 'mutual_info_by_bin_width_' + region + '_' + str(
best_stim) + '.png'
print(dt.datetime.now().isoformat() + ' INFO: ' + 'Saving ' +
filename + '...')
plt.savefig(
os.path.join(image_dir, 'mutual_info_vs_bin_width', x_axis_scale,
filename))
plt.close()
def main():
print(dt.datetime.now().isoformat() + ' INFO: ' +
'Starting main function...')
print(dt.datetime.now().isoformat() + ' INFO: ' +
'Loading data from disc...')
corr_bin_frame = getCorrFrame(args.filename, args.bin_width)
strong_corr_bin_frame = getCorrFrame(args.strong_filename, args.bin_width)
if args.measure_type == 'pairwise':
corr_best_stims, corr_y_lim, corr_max_bin = getBestStimsLims(
corr_bin_frame, strong_corr_bin_frame, 'corr_coef', [-1, 1])
plotRegionalCorrelationHistograms(corr_bin_frame, corr_best_stims,
corr_y_lim, args.use_title,
args.prefix, args.bin_width)
info_best_stims, info_y_lim, info_max_bin = getBestStimsLims(
corr_bin_frame, strong_corr_bin_frame, 'mutual_info_qe')
plotRegionalInformationHistograms(corr_bin_frame, info_best_stims,
info_y_lim, info_max_bin,
args.use_title, args.prefix,
args.bin_width)
symm_best_stims, symm_y_lim, symm_max_bin = getBestStimsLims(
corr_bin_frame, strong_corr_bin_frame, 'symm_unc_qe', [0, 1])
plotRegionalSymmUncHistograms(corr_bin_frame, symm_best_stims,
symm_y_lim, args.use_title, args.prefix,
args.bin_width)
elif args.measure_type == 'single':
firing_best_stims, firing_y_lim, firing_max_bin = getBestStimsLims(
corr_bin_frame, strong_corr_bin_frame, 'firing_rate_mean')
plotRegionalFiringRateHistograms(corr_bin_frame, firing_best_stims,
firing_y_lim, firing_max_bin,
args.use_title, args.prefix,
args.bin_width)
elif args.measure_type == 'correction':
for region in rc.regions:
best_stim = rc.getBestStimFromRegion(correlation_frame, region)
plotCorrections(correlation_frame, region, best_stim,
args.bin_width, args.prefix)
else:
sys.exit(dt.datetime.now().isoformat() + 'ERROR: ' +
'Unrecognised measure type!')
print(dt.datetime.now().isoformat() + ' INFO: ' + 'Done.')
pd.set_option('max_rows', 30)
def getDataFrames(single_file, paired_file, bin_width):
firing_frame = pd.read_csv(os.path.join(csv_dir, args.single_file), usecols=lambda x: x != 'Unnamed: 0')
pairwise_frame = pd.read_csv(os.path.join(csv_dir, args.paired_file), usecols=lambda x: x != 'Unnamed: 0')
firing_frame = firing_frame[firing_frame.bin_width == bin_width]
pairwise_frame = pairwise_frame[pairwise_frame.bin_width == bin_width]
return firing_frame, pairwise_frame
pair = np.array(args.pair)
cell_info, id_adjustor = rc.loadCellInfo(csv_dir)
stim_info = loadmat(os.path.join(mat_dir, 'experiment2stimInfo.mat'))
firing_frame, pairwise_frame = getDataFrames(args.single_file, args.paired_file, args.bin_width)
region = firing_frame.loc[firing_frame.cell_id == pair[0]].region.unique()[0]
best_stim = rc.getBestStimFromRegion(pairwise_frame, region)
pairwise_region_frame = pairwise_frame[(pairwise_frame.region == region) & (pairwise_frame.stim_id == best_stim)]
stim_firing_frame = firing_frame.loc[firing_frame.stim_id == best_stim,:]
trials_info = rc.getStimTimesIds(stim_info, best_stim)
spike_time_dict = rc.loadSpikeTimes(posterior_dir, frontal_dir, pair, id_adjustor)
exp_frame = rc.getExperimentFrame(pair, trials_info, spike_time_dict, cell_info, 1.0)
corr = pairwise_region_frame.loc[(pairwise_region_frame.first_cell_id == pair[0]) & (pairwise_region_frame.second_cell_id == pair[1]), 'corr_coef'].iloc[0]
info = pairwise_region_frame.loc[(pairwise_region_frame.first_cell_id == pair[0]) & (pairwise_region_frame.second_cell_id == pair[1]), 'mutual_info_qe'].iloc[0]
first_firing_rate = stim_firing_frame.loc[firing_frame.cell_id == pair[0], 'firing_rate_mean'].iloc[0]
second_firing_rate = stim_firing_frame.loc[firing_frame.cell_id == pair[1], 'firing_rate_mean'].iloc[0]
geom_mean_firing_rate = np.sqrt(first_firing_rate * second_firing_rate)
first_responses = exp_frame.loc[exp_frame.cell_id == pair[0], 'num_spikes']
second_responses = exp_frame.loc[exp_frame.cell_id == pair[1], 'num_spikes']
plt.figure(figsize=(10,5))