brainbox.population.cca

Functions

bin_spikes_trials

Binarizes the spike times into a raster and assigns a trial number to each bin

fit_cca

Initialize and fit CCA sklearn object

get_cca_projection

Project data into CCA dimensions

get_correlations

param cca:

get_event_bin_indexes

Get the indexes of the bins corresponding to a specific behavioral event within a window

plot_correlations

Correlation vs CCA dimension

plot_pairwise_correlations

Plot CCA correlations for multiple pairs of regions

plot_pairwise_correlations_mult

Plot CCA correlations for multiple pairs of regions, for multiple behavioural events

preprocess

Preprocess neural data for cca analysis with smoothing and pca

shuffle_analysis

Perform CCA on shuffled data

split_by_area

This function converts a matrix of binned spikes into a list of matrices, with the clusters grouped by brain areas

split_timepoints

Assign each time point to testing or training fold

split_trials

Assign each trial to testing or training fold

preprocess(data, smoothing_sd=25, n_pcs=20)[source]

Preprocess neural data for cca analysis with smoothing and pca

Parameters:
  • data (array-like) – array of shape (n_samples, n_features)

  • smoothing_sd (float) – gaussian smoothing kernel standard deviation (ms)

  • n_pcs (int) – number of pca dimensions to retain

Returns:

preprocessed neural data

Return type:

array-like, shape (n_samples, pca_dims)

split_trials(trial_ids, n_splits=5, rng_seed=0)[source]

Assign each trial to testing or training fold

Parameters:
  • trial_ids (array-like)

  • n_splits (int) – one split used for testing; remaining splits used for training

  • rng_seed (int) – set random state for shuffling trials

Returns:

list of dicts of indices with keys train and test

split_timepoints(trial_ids, idxs_trial)[source]

Assign each time point to testing or training fold

Parameters:
  • trial_ids (array-like) – trial id for each timepoint

  • idxs_trial (list) – list of dicts that define which trials are in train or test folds

Returns:

list of dicts that define which time points are in train and test folds

fit_cca(data_0, data_1, n_cca_dims=10)[source]

Initialize and fit CCA sklearn object

Parameters:
  • data_0 (array-like) – shape (n_samples, n_features_0)

  • data_1 (array-like) – shape (n_samples, n_features_1)

  • n_cca_dims (int) – number of CCA dimensions to fit

Returns:

sklearn cca object

get_cca_projection(cca, data_0, data_1)[source]

Project data into CCA dimensions

Parameters:
  • cca

  • data_0

  • data_1

Returns:

tuple; (data_0 projection, data_1 projection)

get_correlations(cca, data_0, data_1)[source]
Parameters:
  • cca

  • data_0

  • data_1

Returns:

shuffle_analysis(data_0, data_1, n_shuffles=100, **cca_kwargs)[source]

Perform CCA on shuffled data

Parameters:
  • data_0

  • data_1

  • n_shuffles

Returns:

plot_correlations(corrs, errors=None, ax=None, **plot_kwargs)[source]

Correlation vs CCA dimension

Parameters:
  • corrs (1-D vector) – correlation values for the CCA dimensions

  • errors – error values

  • ax (matplotlib axis object) – axis to plot on (default None)

Returns:

axis if specified, or plot if axis = None

plot_pairwise_correlations(means, stderrs=None, n_dims=None, region_strs=None, **kwargs)[source]

Plot CCA correlations for multiple pairs of regions

Parameters:
  • means – list of lists; means[i][j] contains the mean corrs between regions i, j

  • stderrs – list of lists; stderrs[i][j] contains std errors of corrs between regions i, j

  • n_dims – number of CCA dimensions to plot

  • region_strs – list of strings identifying each region

  • kwargs – keyword arguments for plot

Returns:

matplotlib figure handle

plot_pairwise_correlations_mult(means, stderrs, colvec, n_dims=None, region_strs=None, **kwargs)[source]

Plot CCA correlations for multiple pairs of regions, for multiple behavioural events

Parameters:
  • means – list of lists; means[k][i][j] contains the mean corrs between regions i, j for behavioral event k

  • stderrs – list of lists; stderrs[k][i][j] contains std errors of corrs between regions i, j for behavioral event k

  • colvec – color vector [must be a better way for this]

  • n_dims – number of CCA dimensions to plot

  • region_strs – list of strings identifying each region

  • kwargs – keyword arguments for plot

Returns:

matplotlib figure handle

bin_spikes_trials(spikes, trials, bin_size=0.01)[source]

Binarizes the spike times into a raster and assigns a trial number to each bin

Parameters:
  • spikes (Bunch) – spikes object

  • trials (Bunch) – trials object

  • bin_size (float) – size, in s, of the bins

Returns:

a matrix (bins, SpikeCounts), and a vector of bins size with trial ID,

and a vector bins size with the time that the bins start

split_by_area(binned_spikes, cl_brainAcronyms, active_clusters, brain_areas)[source]

This function converts a matrix of binned spikes into a list of matrices, with the clusters grouped by brain areas

Parameters:
  • binned_spikes (numpy.ndarray) – binned spike data of shape (n_bins, n_lusters)

  • cl_brainAcronyms (pandas.core.frame.DataFrame) – brain region for each cluster

  • brain_areas (numpy.ndarray) – list of brain areas to select

  • active_clusters (numpy.ndarray) – list of clusterIDs

Returns:

list of numpy.ndarrays of size brain_areas

get_event_bin_indexes(event_times, bin_times, window)[source]

Get the indexes of the bins corresponding to a specific behavioral event within a window

Parameters:
  • event_times (numpy.array) – time series of an event

  • bin_times (numpy.array) – time series pf starting point of bins

  • window (numpy.array) – list of size 2 specifying the window in seconds [-time before, time after]

Returns:

array of indexes