CellScope.cs.Manifold_Fitting_1

CellScope.cs. Manifold_Fitting_1 (fea : np.ndarray | csr_matrix, num_pca : int = 100, num_Selected_Gene : int = 500, knn : int = 20, num_center : int = 0, random_seed : int = 83)

Manifold_Fitting_1 fits the data onto a low-dimensional manifold by identifying significant genes. This process involves dimensionality reduction through PCA [AW10], selecting manifold seeds [RL14], identifying high-confidence cliques, and selecting key genes that define the signal space based on these cliques.

Parameters

• fea (np.ndarray | csr_matrix):

  The feature matrix of shape (n_cell, n_gene), where rows correspond to cells and columns to genes. Supports both dense (NumPy array) and sparse (CSR matrix) formats.

• num_pca (int, optional, default=100):

  The number of principal components to retain after performing PCA on fea. This parameter determines the dimensionality of the PCA-transformed data.

• num_Selected_Gene (int, optional, default=500):

  The number of selected genes representing the signal space, based on the lowest p-values.

• knn (int, optional, default=20):

  The number of nearest neighbors used to compute the local density (rho) and distance to higher density neighbors (delta).

• num_center (int, optional, default=0):

  The number of cluster centers to select. If set to 0, the cluster centers are automatically determined using the findCenters function based on the product of rho and delta.

• random_seed (int, optional, default=83):

  The random seed for ensuring reproducibility of PCA results and Truncated SVD.

Return

• fea_selected (ndarray | csr_matrix):

  A matrix of shape (n_cell, num_Selected_Gene), representing the gene-selected data matrix.

• significant_features_index (ndarray):

  A vector containing the indices of the selected genes, corresponding to the most significant features based on p-values.

• id_max_deltarho (ndarray):

  An array of cluster center indices, representing the cells identified as cluster centers.