neighbors module

velotest.neighbors.cos_directionality(expression, velocity_position, expressions_neighbours)

Calculates the cosine similarity between the velocity of a cell and multiple cells in the neighborhood of the cell. Currently used for 2D velocity vectors.

Parameters:

• expression – position of the cell

• velocity_position – position (x+v) of the tip of the velocity vector

• expressions_neighbours – expressions of the neighbors

Returns:

velotest.neighbors.cos_directionality_batched(expression, velocity_position, expressions_neighbours)

Calculates the cosine similarity between the velocity of a cell and multiple cells in the neighborhood of the cell. Currently used for 2D velocity vectors.

Parameters:

• expression – position of the cell

• velocity_position – position (x+v) of the tip of the velocity vector

• expressions_neighbours – expressions of the neighbors

Returns:

velotest.neighbors.find_neighbors(x, k_neighbors=5, metric='euclidean')

Finds the k nearest neighbors of each cell in the expression matrix.

Parameters:

• x – expression matrix

• k_neighbors – number of neighbors

• metric – metric to use to calculate distances

Returns:

an array of indices of the k nearest neighbors for each cell (#cells, k_neighbors)

velotest.neighbors.find_neighbors_in_direction_of_velocity(Z_expr, Z_velo_position, nn_indices, threshold_degree=22.5) → list

Finds the neighbors of each cell in 2D that are in a 2*threshold_degree cone around the velocity vector.

Parameters:

• Z_expr – 2D embedding of positions

• Z_velo_position – 2D embedding of tip of velocity vectors

• nn_indices – indices of the k nearest neighbors for each cell

• threshold_degree – angle in degrees to define the cone around the velocity vector, e.g. specify 22.5° for a cone of 45°

Returns:

return list of neighbours in direction of velocity for each cell

velotest.neighbors.find_neighbors_in_direction_of_velocity_multiple(Z_expr, Z_velo_position, nn_indices, threshold_degree=22.5) → list

velotest.neighbors.select_neighbors_based_on_cos(Z_expr, Z_velo_position, nn_indices, threshold_degree)

velotest.neighbors.select_neighbors_based_on_cos_batched(Z_expr, Z_velo_position, nn_indices, threshold_degree)