anndict.utils.pca_density_filter_main

anndict.utils.pca_density_filter_main(data, n_components=3, threshold=0.1)

Calculate density contours for PCA-reduced data, return the density of all input data, and identify the indices of the variables that were included in the PCA.

Parameters:

data ndarray | spmatrix

array-like, shape (n_samples, n_features)

n_components int (default: 3)

number of components for PCA to reduce the data to.

threshold float (default: 0.1)

percentile threshold for density values. Points with density above this threshold are considered high-density and will be retained.

Return type:

tuple[ndarray, float, list[int]]

Returns:

A tuple containing: - pca_data: PCA-reduced data (None if all variables are constant). - density: Density values of all the points (None if all variables are constant). - variables_used_in_pca: Indices of unique variables that were included in the PCA (empty list if all variables are constant).