Find outlier cells using a beta distribution — find_outlier

Also inspired by MSKCC SPECTRUM paper. Using the measured pairwise bin distances between cells pairwise_bin_differences(), this function takes the nearest neighbour of each cell and fits a beta distribution. Then using this distribution, it finds outlier cells based on a selected percentile of the distribution (default 99th).

Usage

find_outlier_cells(pairwise_diffs, nn_cells, outlier_percentile = 0.99)

Arguments

nn_cells: dataframe of find_nearest_neighbours()
outlier_percentile: double. Default 0.99. What percentile of the distribution to consider an outlier cell.
cell_diffs: dataframe of differences from pairwise_bin_differences()

Value

NA or tibble of information on cells considered outliers. NA if no outliers found.