allensdk.brain_observatory.receptive_field_analysis.chisquarerf module

allensdk.brain_observatory.receptive_field_analysis.chisquarerf.NLL_to_pvalue(NLLs, log_base=10.0)

allensdk.brain_observatory.receptive_field_analysis.chisquarerf.build_trial_matrix(LSN_template, num_trials, on_off_luminance=(255, 0))

Construct indicator arrays for on/off pixels across trials.

Parameters:

LSN_template : np.ndarray

Dimensions are (nTrials, nYPixels, nXPixels). Luminance values per pixel and trial. The size of the first dimension may be larger than the num_trials argument (in which case only the first num_trials slices will be used) but may not be smaller.

num_trials : int

The number of trials (left-justified) to build indicators for.

on_off_luminance : array-like, optional

The zeroth element is the luminance value of a pixel when on, the first when off. Defaults are [255, 0].

Returns:

trial_mat : np.ndarray

Dimensions are (nYPixels, nXPixels, {on, off}, nTrials). Boolean values indicate that a pixel was on/off on a particular trial.

allensdk.brain_observatory.receptive_field_analysis.chisquarerf.chi_square_binary(events, LSN_template)

allensdk.brain_observatory.receptive_field_analysis.chisquarerf.chi_square_within_mask(exclusion_mask, events_per_pixel, trials_per_pixel)

Determine if cells respond preferentially to on/off pixels in a mask using a chi2 test.

Parameters:

exclusion_mask : np.ndarray

Dimensions are (nYPixels, nXPixels, {on, off}). Integer indicator for INCLUSION (!) of a pixel within the testing region.

events_per_pixel : np.ndarray

Dimensions are (nCells, nYPixels, nXPixels, {on, off}). Integer values are response counts by cell to on/off luminance at each pixel.

trials_per_pixel : np.ndarray

Dimensions are (nYPixels, nXPixels, {on, off}). Integer values are counts of trials where a pixel is on/off.

Returns:

p_vals : np.ndarray

One-dimensional, of length nCells. Float values are p-values for the hypothesis that a given cell has a receptive field within the exclusion mask.

chi : np.ndarray

Dimensions are (nCells, nYPixels, nXPixels, {on, off}). Values (float) are squared residual event counts divided by expected event counts.

allensdk.brain_observatory.receptive_field_analysis.chisquarerf.deinterpolate_RF(rf_map, x_pnts, y_pnts, deg_per_pnt)

Downsample an image

Parameters:

rf_map : np.ndarray

Input image

x_pnts : np.ndarray

Count of sample points along the first (column) axis

y_pnts : np.ndarray

Count of sample points along the zeroth (row) axis

deg_per_pnt : numeric

scale factor

Returns:

sampled_yx : np.ndarray

Downsampled image

allensdk.brain_observatory.receptive_field_analysis.chisquarerf.get_disc_masks(LSN_template, radius=3, on_luminance=255, off_luminance=0)

Obtain an indicator mask surrounding each pixel. The mask is a square, excluding pixels which are coactive on any trial with the main pixel.

Parameters:

LSN_template : np.ndarray

Dimensions are (nTrials, nYPixels, nXPixels). Luminance values per pixel and trial.

radius : int

The base mask will be a box whose sides are 2 * radius + 1 in length.

on_luminance : int, optional

The value of the luminance for on trials. Default is 255

off_luminance : int, optional

The value of the luminance for off trials. Default is 0

Returns:

masks : np.ndarray

Dimensions are (nYPixels, nXPixels, nYPixels, nXPixels). The first 2 dimensions describe the pixel from which the mask was computed. The last 2 serve as the dimensions of the mask images themselves. Masks are binary arrays of type float, with 1 indicating inside, 0 outside.

allensdk.brain_observatory.receptive_field_analysis.chisquarerf.get_events_per_pixel(responses_np, trial_matrix)

Obtain a matrix linking cellular responses to pixel activity.

Parameters:

responses_np : np.ndarray

Dimensions are (nTrials, nCells). Boolean values indicate presence/absence of a response on a given trial.

trial_matrix : np.ndarray

Dimensions are (nYPixels, nXPixels, {on, off}, nTrials). Boolean values indicate that a pixel was on/off on a particular trial.

Returns:

events_per_pixel : np.ndarray

Dimensions are (nCells, nYPixels, nXPixels, {on, off}). Values for each cell, pixel, and on/off state are the sum of events for that cell across all trials where the pixel was in the on/off state.

allensdk.brain_observatory.receptive_field_analysis.chisquarerf.get_expected_events_by_pixel(exclusion_mask, events_per_pixel, trials_per_pixel)

Calculate expected number of events per pixel

Parameters:

exclusion_mask : np.ndarray

Dimensions are (nYPixels, nXPixels, {on, off}). Integer indicator for INCLUSION (!) of a pixel within the testing region.

events_per_pixel : np.ndarray

Dimensions are (nCells, nYPixels, nXPixels, {on, off}). Integer values are response counts by cell to on/off luminance at each pixel.

trials_per_pixel : np.ndarray

Dimensions are (nYPixels, nXPixels, {on, off}). Integer values are counts of trials where a pixel is on/off.

Returns:

np.ndarray :

Dimensions (nCells, nYPixels, nXPixels, {on, off}). Float values are pixelwise counts of events expected if events are evenly distributed in mask across trials.

allensdk.brain_observatory.receptive_field_analysis.chisquarerf.get_peak_significance(chi_square_grid_NLL, LSN_template, alpha=0.05)

allensdk.brain_observatory.receptive_field_analysis.chisquarerf.interpolate_RF(rf_map, deg_per_pnt)

Upsample an image

Parameters:

rf_map : np.ndarray

Input image

deg_per_pnt : numeric

scale factor

Returns:

interpolated : np.ndarray

Upsampled image

allensdk.brain_observatory.receptive_field_analysis.chisquarerf.locate_median(y, x)

allensdk.brain_observatory.receptive_field_analysis.chisquarerf.pvalue_to_NLL(p_values, max_NLL=10.0)

allensdk.brain_observatory.receptive_field_analysis.chisquarerf.smooth_STA(STA, gauss_std=0.75, total_degrees=64)

Smooth an image by convolution with a gaussian kernel

Parameters:

STA : np.ndarray

Input image

gauss_std : numeric, optional

Standard deviation of the gaussian kernel. Will be applied to the upsampled image, so units are visual degrees. Default is 0.75

total_degrees : int, optional

Size in visual degrees of the input image along its zeroth (row) axis. Used to set the scale factor for up/downsampling.

Returns:

STA_smoothed : np.ndarray

Smoothed image