Preprocessing: pp

Contents

Preprocessing: `pp`#

Filtering based on both marker intensity and cell-level observational values/quantiles. Common transforms used for marker intensity values. If backend is set to GPU (and configured), some functions wrap rapids_singlecell functions. Otherwise, same functions wrap scanpy.

Functions#

`pp.set_backend`	Set the backend to use for processing.
`pp.filter_by_obs_count`	Filters cells which belong to a category in `AnnData.obs` with a cell count less than a `min_value` and/or more than a `max_value`.
`pp.filter_by_obs_value`	Filters cells which have a value of a given `AnnData.obs` column less than a `min_value` and/or more than a `max_value`.
`pp.filter_by_obs_quantile`	Filters cells which have a value of a given `AnnData.obs` column less than the `min_quantile` and/or more than the `max_quantile`.
`pp.filter_by_var_value`	Filters cells which have a value of a given `AnnData.var` column less than a `min_value` and/or more than a `max_value`.
`pp.filter_by_var_quantile`	Filters cells which have a value of a given `AnnData.var` column less than the `min_quantile` and/or more than the `max_quantile`.
`pp.fill_na`	Fill NaNs in a given layer or .X with a given value.
`pp.log1p`	Apply log1p transformation (natural log transform with pseudocount) to a given layer or .X.
`pp.arcsinh`	Apply arcsinh transformation with a given cofactor to a given layer or .X.
`pp.zscore`	Apply z-score transformation along the rows of a given layer or .X.
`pp.scale`	Scale columns and rows to have 0 mean and unit variance in a given layer or .X.
`pp.percentile`	Normalise data to the 95th or 99th percentile.
`pp.neighbors`	Compute a neighborhood graph in a given expression or embedding space.
`pp.add_obs_as_var`	Adds .obs features as .var features to an AnnData.
`pp.subset_adata_by_var`	Subsets an AnnData object by var names.
`pp.subset_adata_by_obs_category`	Subsets an AnnData object by obs category or categories.