dotools_py.tl.run_harmony

Contents

dotools_py.tl.run_harmony#

dotools_py.tl.run_harmony(adata, batch_key, use_rep='X_pca', rep_added='X_harmony', random_state=0, use_gpu=False, max_iter_harmony=150, workers=1, **kwargs)[source]#

Run Harmony integration.

This functions runs the Pytorch implementation of Harmony.

Parameters:
adata AnnData

Annotated data matrix.

batch_key str

Key in adata.obs with batch information.

use_rep str (default: 'X_pca')

Representation in adata.obsm that represents the input embedding with rows for cells and columns for embedding coordinates.

rep_added str (default: 'X_harmony')

Name of the adjusted embedding.

random_state int (default: 0)

Seed for random number generator.

use_gpu bool (default: False)

If set to True will use GPU if available.

max_iter_harmony int (default: 150)

Maximum number of iterations for harmony.

workers int (default: 1)

Number of threads to use.

kwargs

Additional arguments pass to harmony.harmonize.

Return type:

None

Returns:

Returns None. A new slot will be set in adata.obsm[rep_added].

Example

>>> import dotools_py as do
>>> adata = do.dt.example_10x_processed()
>>> adata.obsm_keys()
['X_CCA', 'X_pca', 'X_umap']
>>> do.tl.run_harmony(adata, batch_key="batch")
>>> adata.obsm_keys()
['X_CCA', 'X_pca', 'X_umap', 'X_harmony']
Contents