def correct_batch(data: MultimodalData, features: str = None) -> None:
"""Batch correction on data using Location-Scale (L/S) Adjustment method. ([Li-and-Wong03]_, [Li20]_). If L/S adjustment method is used, users must call this function every time before they call the pca function.
Parameters
----------
data: ``pegasusio.MultimodalData``
Annotated data matrix with rows for cells and columns for genes.
features: `str`, optional, default: ``None``
Features to be included in batch correction computation. If ``None``, simply consider all features.
Returns
-------
``None``
Update ``data.X`` by the corrected count matrix.
Examples
--------
>>> pg.correct_batch(data, features = "highly_variable_features")
"""
tot_seconds = 0.0
# estimate adjustment parameters
start = time.perf_counter()
can_correct = estimate_adjustment_matrices(data)
end = time.perf_counter()
tot_seconds += end - start
logger.info("Adjustment parameters are estimated.")
# select dense matrix
keyword = select_features(
data, features=features, standardize=False,
max_value=None) # do not standardize or truncate max_value
logger.info("Features are selected.")
if can_correct:
start = time.perf_counter()
correct_batch_effects(data, keyword, features)
end = time.perf_counter()
tot_seconds += end - start
logger.info(
"Batch correction is finished. Time spent = {:.2f}s.".format(
tot_seconds))
def correct_batch(data: AnnData, features: str = None) -> None:
"""Batch correction on data.
Parameters
----------
data: ``anndata.AnnData``
Annotated data matrix with rows for cells and columns for genes.
features: `str`, optional, default: ``None``
Features to be included in batch correction computation. If ``None``, simply consider all features.
Returns
-------
``None``
Update ``data.X`` by the corrected count matrix.
Examples
--------
>>> pg.correct_batch(adata, features = "highly_variable_features")
"""
tot_seconds = 0.0
# estimate adjustment parameters
start = time.perf_counter()
can_correct = estimate_adjustment_matrices(data)
end = time.perf_counter()
tot_seconds += end - start
logger.info("Adjustment parameters are estimated.")
# select dense matrix
keyword = select_features(data, features)
logger.info("Features are selected.")
if can_correct:
start = time.perf_counter()
correct_batch_effects(data, keyword, features)
end = time.perf_counter()
tot_seconds += end - start
logger.info(
"Batch correction is finished. Time spent = {:.2f}s.".format(tot_seconds)
)
def run_scanorama(
data: MultimodalData,
n_components: int = 50,
features: str = "highly_variable_features",
standardize: bool = True,
max_value: float = 10,
random_state: int = 0,
) -> str:
"""Batch correction using Scanorama.
This is a wrapper of `Scanorama <https://github.com/brianhie/scanorama>`_ package. See [Hie19]_ for details on the algorithm.
Parameters
----------
data: ``MultimodalData``.
Annotated data matrix with rows for cells and columns for genes.
n_components: ``int``, optional default: ``50``.
Number of integrated embedding components to keep. This sets Scanorama's dimred parameter.
features: ``str``, optional, default: ``"highly_variable_features"``.
Keyword in ``data.var`` to specify features used for Scanorama.
standardize: ``bool``, optional, default: ``True``.
Whether to scale the data to unit variance and zero mean.
max_value: ``float``, optional, default: ``10``.
The threshold to truncate data after scaling. If ``None``, do not truncate.
random_state: ``int``, optional, default: ``0``.
Seed for random number generator.
Returns
-------
out_rep: ``str``
The keyword in ``data.obsm`` referring to the embedding calculated by Scanorama algorithm. out_rep is always equal to "scanorama"
Update ``data.obsm``:
* ``data.obsm['X_scanorama']``: The embedding calculated by Scanorama algorithm.
Examples
--------
>>> pg.run_scanorama(data, random_state = 25)
"""
if not is_categorical_dtype(data.obs['Channel']):
data.obs['Channel'] = pd.Categorical(data.obs['Channel'])
if data.obs['Channel'].cat.categories.size == 1:
logger.warning("Warning: data only contains 1 channel. Cannot apply Scanorama!")
return 'pca'
try:
from scanorama import integrate
except ImportError as e:
print(f"ERROR: {e}")
print("ERROR: Need Scanorama! Try 'pip install scanorama'.")
import sys
sys.exit(-1)
logger.info("Start integration using Scanorama.")
rep = 'scanorama'
keyword = select_features(data, features=features, standardize=standardize, max_value=max_value, use_cache=False)
X = data.uns[keyword]
datasets = []
for channel in data.obs['Channel'].cat.categories:
idx = (data.obs['Channel'] == channel).values
assert idx.sum() > 0
datasets.append(X[idx, :])
genes_list = [[str(i) for i in range(X.shape[1])]] * data.obs['Channel'].cat.categories.size
integrated, genes = integrate(datasets, genes_list, dimred = n_components, seed = random_state)
data.obsm[f'X_{rep}'] = np.concatenate(integrated, axis = 0)
return rep