CellScope.cs.Manifold_Fitting_2

CellScope.cs. Manifold_Fitting_2 (fea_selected : np.ndarray, num_neighbor : int = 5, fitting_prop : float = 0.05, coeff : float = 0.1, op_Outlier : bool = False)

Manifold_Fitting_2 reduces the impact of ambiguous manifold boundaries on downstream analysis by pulling low-density cells closer to higher-density ones. This process involves identifying low-density cells, adjusting potential outliers, and further fitting a subset of cells to minimize noise.

Parameters

• fea_selected (np.ndarray):

  The selected feature matrix from the first step of Manifold Fitting, with dimensions (n_cell, num_Selected_Gene).

• num_neighbor (int, optional, default=5):

  The number of nearest neighbors to consider when calculating local density in the second manifold fitting step.

• fitting_prop (float, optional, default=0.05):

  The proportion of cells selected for the second manifold fitting step, representing a fraction of the total cells. It must be a float between 0 and 1.

• coeff (float, optional, default=0.1):

  The coefficient used to update the fitting cells by averaging with their nearest neighbors. It must be a float between 0 and 1.

• op_Outlier (bool, optional, default=False):

  Specifies whether to identify and remove outliers among the fitting cells during the second manifold fitting step.

Return

• fea_selected (np.ndarray):

  The updated feature matrix after the second step of manifold fitting, with dimensions (n_remaining_cells, num_Selected_Gene), where n_remaining_cells is the number of cells left after fitting and optional outlier removal.

• fitting_index (ndarray):

  The indices of the cells selected for the second manifold fitting step, representing the most significant fitting cells.

• index (ndarray):

  The indices of the remaining cells after removing outliers, representing the final set of cells included in the adjusted feature matrix.