dynamo.vf.rank_divergence_genes

dynamo.vf.rank_divergence_genes(adata, jkey='jacobian_pca', genes=None, prefix_store='rank_div_gene', **kwargs)[source]
Rank genes based on their diagonal Jacobian for each cell group.

Be aware that this ‘divergence’ refers to the diagonal elements of a gene-wise Jacobian, rather than its trace, which is the common definition of the divergence.

Run .vf.jacobian and set store_in_adata=True before using this function.

Parameters:
  • adata (AnnData) – AnnData object that contains the reconstructed vector field in the .uns attribute.

  • jkey (str) – The key in .uns of the cell-wise Jacobian matrix.

  • genes (Optional[List]) – A list of names for genes of interest.

  • prefix_store (str) – The prefix added to the key for storing the returned ranking info in adata.

  • kwargs

    additional keys that will be passed to the rank_genes function. It will accept the following arguments: group: str or None (default: None)

    The cell group that speed ranking will be grouped-by.

    genes: list or None (default: None)

    The gene list that speed will be ranked. If provided, they must overlap the dynamics genes.

    abs: bool (default: False)

    When pooling the values in the array (see below), whether to take the absolute values.

    normalize: bool (default: False)

    Whether normalize the array across all cells first, if the array is 2d.

    fcn_pool: callable (default: numpy.mean(x, axis=0))

    The function used to pool values in the to-be-ranked array if the array is 2d.

    output_values: bool (default: False)

    Whether output the values along with the rankings.

Return type:

DataFrame

Returns:

AnnData object which has the rank dictionary for diagonal jacobians in .uns.