from __future__ import annotations
from collections import defaultdict
from itertools import chain
from typing import Any
import numpy as np
import pandas as pd
from _collections import OrderedDict
from anndata import AnnData
from category_encoders import CountEncoder, HashingEncoder
from rich import print
from rich.progress import BarColumn, Progress
from sklearn.preprocessing import LabelEncoder, OneHotEncoder
from ehrapy import logging as logg
from ehrapy.anndata.anndata_ext import _update_uns
multi_encoding_modes = {"hash_encoding"}
available_encodings = {"one_hot_encoding", "label_encoding", "count_encoding", *multi_encoding_modes}
[docs]def encode(
data: AnnData,
autodetect: bool | dict = False,
encodings: dict[str, dict[str, list[str]]] | dict[str, list[str]] | str | None = None,
) -> AnnData:
"""Encode categoricals of an :class:`~anndata.AnnData` object.
Categorical values could be either passed via parameters or are autodetected on the fly.
The categorical values are also stored in obs and uns (for keeping the original, unencoded values).
The current encoding modes for each variable are also stored in uns (`var_to_encoding` key).
Variable names in var are updated according to the encoding modes used. A variable name starting with `ehrapycat_`
indicates an encoded column (or part of it).
Autodetect mode:
This can be used for convenience and when there are many columns that need to be encoded. Note that missing values do not influence the result.
By using this mode, every column that contains non-numerical values is encoded. In addition, every binary
column will be encoded too. These are those columns which contain only 1's and 0's (could be either integers or floats).
Available encodings are:
1. one_hot_encoding (https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.OneHotEncoder.html)
2. label_encoding (https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.LabelEncoder.html)
3. count_encoding (https://contrib.scikit-learn.org/category_encoders/count.html)
4. hash_encoding (https://contrib.scikit-learn.org/category_encoders/hashing.html)
Args:
data: A :class:`~anndata.AnnData` object.
autodetect: Whether to autodetect categorical values that will be encoded.
encodings: Only needed if autodetect set to False.
A dict containing the encoding mode and categorical name for the respective column
or the specified encoding that will be applied to all columns.
Returns:
An :class:`~anndata.AnnData` object with the encoded values in X.
Examples:
>>> import ehrapy as ep
>>> adata = ep.dt.mimic_2()
>>> # encode all autodetected (non numerical) columns using label encoding
>>> adata_encoded = ep.pp.encode(adata, autodetect=True)
# Example using autodetect with non-default encoding mode:
>>> import ehrapy as ep
>>> adata = ep.dt.mimic_2()
>>> # encode all autodetected (non numerical) columns using one hot encoding
>>> adata_encoded = ep.pp.encode(adata, autodetect=True, encodings='one_hot_encoding')
# Example using custom encodings per columns:
>>> import ehrapy as ep
>>> adata = ep.dt.mimic_2()
>>> # encode col1 and col2 using label encoding and encode col3 using one hot encoding
>>> adata_encoded = ep.pp.encode(adata,
>>> autodetect=False,
>>> encodings={'label_encoding': ['col1', 'col2'], 'one_hot_encoding': ['col3']})
"""
if isinstance(data, AnnData):
# basic type checking for passed parameters when encoding a single AnnData object
_check_anndata_input_type(autodetect, encodings)
return _encode(adata=data, autodetect=autodetect, encodings=encodings)
else:
raise ValueError(f"Cannot encode object of type {type(data)}. Can only encode AnnData objects!")
[docs]def undo_encoding(
data: AnnData,
columns: str = "all",
) -> AnnData | None:
"""Undo the current encodings applied to all columns in X.
This currently resets the AnnData object to its initial state.
Args:
data: The :class:`~anndata.AnnData` object
columns: The names of the columns to reset encoding for. Defaults to all columns.
Returns:
A (partially) encoding reset :class:`~anndata.AnnData`
Examples:
>>> import ehrapy as ep
>>> # adata_encoded is an encoded AnnData object
>>> adata_undone = ep.pp.encode.undo_encoding(adata_encoded)
"""
if isinstance(data, AnnData):
return _undo_encoding(data, columns)
else:
print(f"[b red]Cannot decode object of type {type(data)}. Can only decode AnnData objects!")
raise ValueError
return None
def _encode(
adata: AnnData,
autodetect: bool | dict = False,
encodings: dict[str, dict[str, list[str]]] | dict[str, list[str]] | str = None,
) -> AnnData | None:
"""Encode the initial read AnnData object. Categorical values could be either passed via parameters or autodetected.
Label encodes by default which is used to save initially unencoded AnnData objects.
Args:
adata: The inital AnnData object parsed by the :class: DataReader
autodetect: Autodetection of categorical values (default: False)
encodings: Only needed if autodetect set to False. A dict containing the categorical name
and the encoding mode for the respective column
Returns:
An :class:`~anndata.AnnData` object with the encoded values in X
"""
# check whether original layer exists; if not add it
if "original" not in adata.layers.keys():
adata.layers["original"] = adata.X.copy()
# autodetect categorical values, which could lead to more categoricals
if autodetect:
if "var_to_encoding" in adata.uns.keys():
print(
"[bold yellow]The current AnnData object has been already encoded. Returning original AnnData object!"
)
return adata
categoricals_names = adata.uns["non_numerical_columns"]
# no columns were detected, that would require an encoding (e.g. non numerical columns)
if not categoricals_names:
print("[bold yellow]Detected no columns that need to be encoded. Leaving passed AnnData object unchanged.")
return adata
# copy uns so it can be used in encoding process without mutating the original anndata object
orig_uns_copy = adata.uns.copy()
_add_categoricals_to_uns(adata, orig_uns_copy, categoricals_names)
encoded_x = None
encoded_var_names = adata.var_names.to_list()
encode_mode = encodings if encodings is not None else "label_encoding"
if encode_mode not in available_encodings - multi_encoding_modes:
print(
f"Unknown encoding mode {encode_mode}. Please provide one of the following encoding modes:\n"
f"{available_encodings - multi_encoding_modes}"
)
raise ValueError
single_encode_mode_switcher = {
"one_hot_encoding": _one_hot_encoding,
"label_encoding": _label_encoding,
"count_encoding": _count_encoding,
}
with Progress(
"[progress.description]{task.description}",
BarColumn(),
"[progress.percentage]{task.percentage:>3.0f}%",
refresh_per_second=1500,
) as progress:
task = progress.add_task(f"[red]Running {encode_mode} on detected columns ...", total=1)
# encode using the desired mode
encoded_x, encoded_var_names = single_encode_mode_switcher[encode_mode]( # type: ignore
adata,
encoded_x,
orig_uns_copy,
encoded_var_names,
categoricals_names,
progress,
task,
)
progress.update(task, description="Updating layer originals ...")
# update layer content with the latest categorical encoding and the old other values
logg.info("Encoding strings in X to save to .h5ad. Loading the file will reverse the encoding.")
updated_layer = _update_layer_after_encoding(
adata.layers["original"],
encoded_x,
encoded_var_names,
adata.var_names.to_list(),
categoricals_names,
)
progress.update(task, description=f"[bold blue]Finished {encode_mode} of autodetected columns.")
encoded_ann_data = AnnData(
encoded_x,
obs=adata.obs.copy(),
var=dict(var_names=encoded_var_names),
uns=orig_uns_copy,
layers={"original": updated_layer},
dtype=np.float32,
)
encoded_ann_data.uns["var_to_encoding"] = {categorical: encode_mode for categorical in categoricals_names}
encoded_ann_data.uns["encoding_to_var"] = {encode_mode: categoricals_names}
encoded_ann_data.uns["numerical_columns"] = adata.uns["numerical_columns"].copy()
encoded_ann_data.uns["non_numerical_columns"] = []
encoded_ann_data.uns["encoded_non_numerical_columns"] = [
column for column in encoded_ann_data.var_names if column.startswith("ehrapycat_")
]
_add_categoricals_to_obs(adata, encoded_ann_data, categoricals_names)
# user passed categorical values with encoding mode for each of them
else:
# Required since this would be deleted through side references
non_numericals = adata.uns["non_numerical_columns"].copy()
# reencode data
if "var_to_encoding" in adata.uns.keys():
encodings = _reorder_encodings(adata, encodings) # type: ignore
adata = _undo_encoding(adata, "all")
adata.uns["non_numerical_columns"] = non_numericals
# are all specified encodings valid?
for encoding_mode in encodings.keys(): # type: ignore
if encoding_mode not in available_encodings:
print(
f"Unknown encoding mode {encoding_mode}. Please provide one of the following encoding modes:\n"
f"{available_encodings}"
)
raise ValueError
adata.uns["encoding_to_var"] = encodings
categoricals_not_flat = list(chain(*encodings.values())) # type: ignore
# this is needed since multi column encoding will get passed a list of list instead of a flat list
categoricals = list(
chain(
*(
_categoricals if isinstance(_categoricals, list) else (_categoricals,)
for _categoricals in categoricals_not_flat
)
)
)
# ensure no categorical column gets encoded twice
if len(categoricals) != len(set(categoricals)):
raise ValueError(
"The categorical column names given contain at least one duplicate column. "
"Check the column names to ensure that no column is encoded twice!"
)
elif any(cat in adata.uns["numerical_columns"] for cat in categoricals):
print(
"[bold yellow]At least one of passed column names seems to have numerical dtype. In general it is not recommended "
"to encode numerical columns!"
)
orig_uns_copy = adata.uns.copy()
_add_categoricals_to_uns(adata, orig_uns_copy, categoricals)
var_to_encoding = {} if "var_to_encoding" not in adata.uns.keys() else adata.uns["var_to_encoding"]
encoded_x = None
encoded_var_names = adata.var_names.to_list()
with Progress(
"[progress.description]{task.description}",
BarColumn(),
"[progress.percentage]{task.percentage:>3.0f}%",
refresh_per_second=1500,
) as progress:
for encoding_mode in encodings.keys(): # type: ignore
task = progress.add_task(f"[red]Setting up {encoding_mode}", total=1)
encode_mode_switcher = {
"one_hot_encoding": _one_hot_encoding,
"label_encoding": _label_encoding,
"count_encoding": _count_encoding,
"hash_encoding": _hash_encoding,
}
progress.update(task, description=f"Running {encoding_mode} ...")
# perform the actual encoding
encoded_x, encoded_var_names = encode_mode_switcher[encoding_mode](
adata, encoded_x, orig_uns_copy, encoded_var_names, encodings[encoding_mode], progress, task # type: ignore
)
# update encoding history in uns
for categorical in encodings[encoding_mode]: # type: ignore
# multi column encoding modes -> multiple encoded columns
if isinstance(categorical, list):
for column_name in categorical:
var_to_encoding[column_name] = encoding_mode
else:
var_to_encoding[categorical] = encoding_mode
# update original layer content with the new categorical encoding and the old other values
updated_layer = _update_layer_after_encoding(
adata.layers["original"],
encoded_x,
encoded_var_names,
adata.var_names.to_list(),
categoricals,
)
try:
encoded_ann_data = AnnData(
encoded_x,
obs=adata.obs.copy(),
var=dict(var_names=encoded_var_names),
uns=orig_uns_copy,
layers={"original": updated_layer},
dtype=np.float32,
)
# update current encodings in uns
encoded_ann_data.uns["var_to_encoding"] = var_to_encoding
# if the user did not pass every non numerical column for encoding, a Anndata object cannot be created
except ValueError:
print(
"[bold red]Creation of AnnData object failed. Ensure that you passed all non numerical, "
"categorical values for encoding!"
)
raise AnnDataCreationError
updated_num_uns, updated_non_num_uns, _ = _update_uns(adata, categoricals)
encoded_ann_data.uns["numerical_columns"] = updated_num_uns
encoded_ann_data.uns["non_numerical_columns"] = updated_non_num_uns
encoded_ann_data.uns["encoded_non_numerical_columns"] = [
column for column in encoded_ann_data.var_names if column.startswith("ehrapycat_")
]
_add_categoricals_to_obs(adata, encoded_ann_data, categoricals)
logg.debug("Successfully encoded the AnnData object.")
return encoded_ann_data
def _one_hot_encoding(
adata: AnnData,
X: np.ndarray | None,
uns: dict[str, Any],
var_names: list[str],
categories: list[str],
progress: Progress,
task,
) -> tuple[np.ndarray, list[str]]:
"""Encode categorical columns using one hot encoding.
Args:
adata: The current AnnData object
X: Current (encoded) X
uns: A copy of the original uns
var_names: Var names of current AnnData object
categories: The name of the categorical columns to be encoded
Returns:
Encoded new X and the corresponding new var names
"""
original_values = _initial_encoding(uns, categories)
progress.update(task, description="[blue]Running one hot encoding on passed columns ...")
encoder = OneHotEncoder(handle_unknown="ignore", sparse=False).fit(original_values)
categorical_prefixes = [
f"ehrapycat_{category}_{str(suffix).strip()}"
for idx, category in enumerate(categories)
for suffix in encoder.categories_[idx]
]
transformed = encoder.transform(original_values)
# X is None if this is the first encoding "round" -> take the former X
if X is None:
X = adata.X
progress.advance(task, 1)
progress.update(task, description="[blue]Updating X and var ...")
temp_x, temp_var_names = _update_encoded_data(X, transformed, var_names, categorical_prefixes, categories)
progress.update(task, description="[blue]Finished one hot encoding.")
return temp_x, temp_var_names
def _label_encoding(
adata: AnnData,
X: np.ndarray | None,
uns: dict[str, Any],
var_names: list[str],
categoricals: list[str],
progress: Progress,
task,
) -> tuple[np.ndarray, list[str]]:
"""Encode categorical columns using label encoding.
Args:
adata: The current AnnData object
X: Current (encoded) X
uns: A copy of the original uns
var_names: Var names of current AnnData object
categoricals: The name of the categorical columns, that need to be encoded
Returns:
Encoded new X and the corresponding new var names
"""
original_values = _initial_encoding(uns, categoricals)
# label encoding expects input array to be 1D, so iterate over all columns and encode them one by one
for idx in range(original_values.shape[1]):
progress.update(task, description=f"[blue]Running label encoding on column {categoricals[idx]} ...")
label_encoder = LabelEncoder()
row_vec = original_values[:, idx : idx + 1].ravel() # type: ignore
label_encoder.fit(row_vec)
transformed = label_encoder.transform(row_vec)
# need a column vector instead of row vector
original_values[:, idx : idx + 1] = transformed[..., None]
progress.advance(task, 1 / len(categoricals))
category_prefixes = [f"ehrapycat_{categorical}" for categorical in categoricals]
# X is None if this is the first encoding "round" -> take the former X
if X is None:
X = adata.X
progress.update(task, description="[blue]Updating X and var ...")
temp_x, temp_var_names = _update_encoded_data(X, original_values, var_names, category_prefixes, categoricals)
progress.update(task, description="[blue]Finished label encoding.")
return temp_x, temp_var_names
def _count_encoding(
adata: AnnData,
X: np.ndarray | None,
uns: dict[str, Any],
var_names: list[str],
categoricals: list[str],
progress: Progress,
task,
) -> tuple[np.ndarray, list[str]]:
"""Encode categorical column using count encoding.
Args:
adata: The current AnnData object
X: Current (encoded) X
uns: A copy of the original uns
var_names: Var names of current AnnData object
categoricals: The name of the categorical columns, that need to be encoded
Returns:
Encoded new X and the corresponding new var names
"""
original_values = _initial_encoding(uns, categoricals)
progress.update(task, description="[blue]Running count encoding encoding on passed columns ...")
# returns a pandas dataframe per default, but numpy array is needed
count_encoder = CountEncoder(return_df=False)
count_encoder.fit(original_values)
category_prefix = [f"ehrapycat_{categorical}" for categorical in categoricals]
transformed = count_encoder.transform(original_values)
# X is None if this is the first encoding "round" -> take the former X
if X is None:
X = adata.X # noqa: N806
progress.advance(task, 1)
progress.update(task, description="[blue]Updating X and var ...")
temp_x, temp_var_names = _update_encoded_data(X, transformed, var_names, category_prefix, categoricals)
progress.update(task, description="[blue]Finished count encoding.")
return temp_x, temp_var_names
def _hash_encoding(
adata: AnnData,
X: np.ndarray | None,
uns: dict[str, Any],
var_names: list[str],
categories: list[list[str]],
progress: Progress,
task,
) -> tuple[np.ndarray, list[str]]:
"""Encode categorical columns using hash encoding.
Args:
adata: The current AnnData object
X: Current (encoded) X
uns: A copy of the original uns
var_names: Var names of current AnnData object
categories: The name of the categorical columns to be encoded
Returns:
Encoded new X and the corresponding new var names
"""
transformed_all, encoded_var_names = None, []
for idx, multi_columns in enumerate(categories):
progress.update(task, description=f"Running hash encoding on {idx + 1}. list ...")
original_values = _initial_encoding(uns, multi_columns)
encoder = HashingEncoder(return_df=False, n_components=8).fit(original_values)
encoded_var_names += [f"ehrapycat_hash_{multi_columns[0]}" for _ in range(8)]
transformed = encoder.transform(original_values)
transformed_all = np.hstack((transformed_all, transformed)) if transformed_all is not None else transformed
progress.advance(task, 1 / len(categories))
# X is None if this is the first encoding "round" -> take the former X
if X is None:
X = adata.X
progress.update(task, description="[blue]Updating X and var ...")
temp_x, temp_var_names = _update_multi_encoded_data(
X, transformed_all, var_names, encoded_var_names, sum(categories, [])
)
if temp_x.shape[1] != len(temp_var_names):
raise HashEncodingError(
"Hash encoding of input data failed. Note that hash encoding is not "
"suitable for datasets with low number of data points and low cardinality!"
)
progress.update(task, description="[blue]Finished hash encoding.")
return temp_x, temp_var_names
def _update_layer_after_encoding(
old_layer: np.ndarray,
new_x: np.ndarray,
new_var_names: list[str],
old_var_names: list[str],
categories: list[str],
) -> np.ndarray:
"""Update the original layer containing the initial non categorical values and the latest encoded categoricals.
Args:
old_layer: The previous "original" layer
new_x: The new encoded X
new_var_names: The new encoded var names
old_var_names: The previous var names
categories: All previous categorical names
Returns
A Numpy array containing all numericals together with all encoded categoricals.
"""
try:
# get the index of the first column of the new encoded X, that does not store an encoded categorical
new_cat_stop_index = next(i for i in range(len(new_var_names)) if not new_var_names[i].startswith("ehrapycat"))
# get the index of the first column of the old encoded X, that does not store an encoded categorical
old_cat_stop_index = next(i for i in range(len(old_var_names)) if not old_var_names[i].startswith("ehrapycat"))
# when there are only encoded columns, simply return a copy of the new X, since to originals will be kept in the layer
except StopIteration:
return new_x.copy().astype("float32")
# keep track of all indices with original value columns, that are (and were) not encoded
idx_list = []
for idx, col_name in enumerate(old_var_names[old_cat_stop_index:]):
# this case is needed when there are one or more numerical (but categorical) columns that was not encoded yet
if col_name not in categories:
idx_list.append(idx + old_cat_stop_index)
# slice old original layer using the selector
old_layer_view = old_layer[:, idx_list]
# get all encoded categoricals of X
encoded_categoricals = new_x[:, :new_cat_stop_index]
# horizontally stack all encoded categoricals and the remaining "old original values"
updated_layer = np.hstack((encoded_categoricals, old_layer_view))
try:
logg.info("Updated the original layer after encoding.")
return updated_layer.astype("float32")
except ValueError as e:
raise ValueError("Ensure that all columns which require encoding are being encoded.") from e
def _update_multi_encoded_data(
X: np.ndarray,
transformed: np.ndarray,
var_names: list[str],
encoded_var_names: list[str],
categoricals: list[str],
) -> tuple[np.ndarray, list[str]]:
"""Update X and var_names after applying multi column encoding modes to some columns
Args:
X: Current (former) X
transformed: The encoded (transformed) categorical columns
var_names: Var names of current AnnData object
encoded_var_names: The name(s) of the encoded column(s)
categoricals: The categorical values that were encoded recently
Returns:
Encoded new X and the corresponding new var names
"""
idx = []
for pos, name in enumerate(var_names):
if name in categoricals:
idx.append(pos)
# delete the original categorical column
del_cat_column_x = np.delete(X, list(idx), 1)
# create the new, encoded X
temp_x = np.hstack((transformed, del_cat_column_x))
# delete old categorical name
var_names = [col_name for col_idx, col_name in enumerate(var_names) if col_idx not in idx]
temp_var_names = encoded_var_names + var_names
logg.debug("Updated `X` and `var_names` after applying multi column encoding modes.")
return temp_x, temp_var_names
def _update_encoded_data(
X: np.ndarray,
transformed: np.ndarray,
var_names: list[str],
categorical_prefixes: list[str],
categoricals: list[str],
) -> tuple[np.ndarray, list[str]]:
"""Update X and var_names after each encoding.
Args:
X: Current (former) X
transformed: The encoded (transformed) categorical column
var_names: Var names of current AnnData object
categorical_prefixes: The name(s) of the encoded column(s)
categoricals: The categorical values that were encoded recently
Returns:
Encoded new X and the corresponding new var names
"""
idx = _get_categoricals_old_indices(var_names, categoricals)
# delete the original categorical column
del_cat_column_x = np.delete(X, list(idx), 1)
# create the new, encoded X
temp_x = np.hstack((transformed, del_cat_column_x))
# delete old categorical name
var_names = [col_name for col_idx, col_name in enumerate(var_names) if col_idx not in idx]
temp_var_names = categorical_prefixes + var_names
logg.debug("Updated `X` and `var_names` after each encoding.")
return temp_x, temp_var_names
def _initial_encoding(
uns: dict[str, Any],
categoricals: list[str],
) -> np.ndarray:
"""Get all original values for all categoricals that need to be encoded (again).
Args:
uns: A copy of the original AnnData object's uns
categoricals: All categoricals that need to be encoded
Returns:
Numpy array of all original categorial values
"""
uns_: dict[str, np.ndarray] = uns
# create numpy array from all original categorical values, that will be encoded (again)
array = np.array(
[uns_["original_values_categoricals"][categoricals[i]].ravel() for i in range(len(categoricals))]
).transpose()
return array
def _undo_encoding(
adata: AnnData,
columns: str = "all",
suppress_warning: bool = False,
) -> AnnData | None:
"""
Undo the current encodings applied to all columns in X. This currently resets the AnnData object to its initial state.
Args:
adata: The AnnData object
columns: The names of the columns to reset encoding for. Defaults to all columns. This resets the AnnData object to its initial state.
suppress_warning: Whether warnings should be suppressed or not.
Returns:
A (partially) encoding reset AnnData object
"""
if "var_to_encoding" not in adata.uns.keys():
if not suppress_warning:
print("[bold yellow]Calling undo_encoding on unencoded AnnData object.")
return None
# get all encoded variables
encoded_categoricals = list(adata.uns["original_values_categoricals"].keys())
# get all columns that should be stored in obs only
columns_obs_only = [
column_name for column_name in list(adata.obs.columns) if column_name not in encoded_categoricals
]
if columns == "all":
categoricals = list(adata.uns["original_values_categoricals"].keys())
else:
print("[bold yellow]Currently, one can only reset encodings for all columns! [bold red]Aborting...")
return None
transformed = _initial_encoding(adata.uns, categoricals)
temp_x, temp_var_names = _delete_all_encodings(adata)
new_x = np.hstack((transformed, temp_x)) if temp_x is not None else transformed
new_var_names = categoricals + temp_var_names if temp_var_names is not None else categoricals
# only keep columns in obs that were stored in obs only -> delete every encoded column from obs
new_obs = adata.obs[columns_obs_only]
uns = OrderedDict()
# reset uns and keep numerical/non-numerical columns
num_vars, non_num_vars = adata.uns["numerical_columns"], adata.uns["non_numerical_columns"]
for cat in categoricals:
original_values = adata.uns["original_values_categoricals"][cat]
type_first_nan = original_values[np.where(original_values != np.nan)][0]
if isinstance(type_first_nan, (int, float, complex)) and not isinstance(type_first_nan, bool):
num_vars.append(cat)
else:
non_num_vars.append(cat)
uns["numerical_columns"] = num_vars
uns["non_numerical_columns"] = non_num_vars
logg.debug("Encoding of the AnnData object was reset.")
return AnnData(
new_x,
obs=new_obs,
var=pd.DataFrame(index=new_var_names),
uns=uns,
dtype=np.dtype(object),
layers={"original": new_x.copy()},
)
def _delete_all_encodings(adata: AnnData) -> tuple[np.ndarray | None, list | None]:
"""Delete all encoded columns and keep track of their indices.
Args:
adata: The AnnData object to operate on
Returns:
A temporary X were all encoded columns are deleted and all var_names of unencoded columns.
"""
var_names = list(adata.var_names)
if adata.X is not None and var_names is not None:
idx = 0
for var in var_names:
if not var.startswith("ehrapycat"):
break
idx += 1
# case: only encoded columns were found
if idx == len(var_names):
return None, None
# don't need to consider case when no encoded columns are there, since undo_encoding would not run anyways in this case
logg.info("All encoded columns of the AnnData object were deleted.")
return adata.X[:, idx:].copy(), var_names[idx:]
return None, None
def _reorder_encodings(adata: AnnData, new_encodings: dict[str, list[list[str]] | list[str]]):
"""Reorder the encodings and update which column will be encoded using which mode (with which columns in case of multi column encoding modes).
Args:
adata: The AnnData object to be reencoded
new_encodings: The new encodings passed by the user (might affect encoded as well as previously non encoded columns)
Returns:
An updated encoding scheme
"""
flattened_modes: list[list[str] | str] = sum(new_encodings.values(), []) # type: ignore
latest_encoded_columns = list(chain(*(i if isinstance(i, list) else (i,) for i in flattened_modes)))
# check for duplicates and raise an error if any
if len(set(latest_encoded_columns)) != len(latest_encoded_columns):
print(
"[bold red]Reencoding AnnData object failed. You have at least one duplicate in your encodings. A column "
"cannot be encoded at the same time using different encoding modes!"
)
raise DuplicateColumnEncodingError
old_encode_mode = adata.uns["var_to_encoding"]
for categorical in latest_encoded_columns:
encode_mode = old_encode_mode.get(categorical)
# if None, this categorical has not been encoded before but will be encoded now for the first time
# multi column encoding mode
if encode_mode in multi_encoding_modes:
encoded_categoricals_with_mode = adata.uns["encoding_to_var"][encode_mode]
_found = False
for column_list in encoded_categoricals_with_mode:
for column_name in column_list:
if column_name == categorical:
column_list.remove(column_name)
_found = True
break
# a categorical can only be encoded once and therefore found once
if _found:
break
# filter all lists that are now empty since all variables will be reencoded from this list
updated_multi_list = filter(None, encoded_categoricals_with_mode)
# if no columns remain that will be encoded with this encode mode, delete this mode from modes as well
if not list(updated_multi_list):
del adata.uns["encoding_to_var"][encode_mode]
# single column encoding mode
elif encode_mode in available_encodings:
encoded_categoricals_with_mode = adata.uns["encoding_to_var"][encode_mode]
for ind, column_name in enumerate(encoded_categoricals_with_mode):
if column_name == categorical:
del encoded_categoricals_with_mode[ind]
break
# if encoding mode is
if not encoded_categoricals_with_mode:
del adata.uns["encoding_to_var"][encode_mode]
logg.info("Re-encoded the AnnData object.")
# return updated encodings
return _update_new_encode_modes(new_encodings, adata.uns["encoding_to_var"])
def _update_new_encode_modes(
new_encodings: dict[str, list[list[str]] | list[str]],
filtered_old_encodings: dict[str, list[list[str]] | list[str]],
):
"""Update the encoding scheme.
If the encoding mode exists in the filtered old encodings, append all values (columns that should be encoded using this mode) to this key.
If not, defaultdict ensures that no KeyError will be raised and the values are simply appended to the default value ([]).
Args:
new_encodings: The new encoding modes passed by the user; basically what will be passed for encodings when calling the encode API
filtered_old_encodings: The old encoding modes, but with all columns removed that will be reencoded
Returns:
The updated encoding scheme
"""
updated_encodings = defaultdict(list) # type: ignore
for k, v in chain(new_encodings.items(), filtered_old_encodings.items()):
updated_encodings[k] += v
logg.debug("Updated the encoding scheme.")
return dict(updated_encodings)
def _get_categoricals_old_indices(old_var_names: list[str], encoded_categories: list[str]) -> set[int]:
"""Get indices of every (possibly encoded) categorical column belonging to a newly encoded categorical value.
Args:
old_var_names: Former variables names
encoded_categories: Already encoded categories
Returns:
Set of all indices of formerly encoded categories belonging to a newly encoded categorical value.
"""
idx_list = set()
category_set = set(encoded_categories)
for idx, old_var_name in enumerate(old_var_names):
# if the old variable was previously unencoded (only the case for numerical categoricals)
if old_var_name in category_set:
idx_list.add(idx)
# if the old variable was already encoded
elif old_var_name.startswith("ehrapycat_"):
if any(old_var_name[10:].startswith(category) for category in category_set):
idx_list.add(idx)
return idx_list
def _add_categoricals_to_obs(original: AnnData, new: AnnData, categorical_names: list[str]) -> None:
"""Add the original categorical values to obs.
Args:
original: The original AnnData object
new: The new AnnData object
categorical_names: Name of each categorical column
"""
for idx, var_name in enumerate(original.var_names):
if var_name in new.obs.columns:
continue
elif var_name in categorical_names:
new.obs[var_name] = original.X[::, idx : idx + 1].flatten()
# note: this will count binary columns (0 and 1 only) as well
# needed for writing to .h5ad files
if set(pd.unique(new.obs[var_name])).issubset({False, True, np.NaN}):
new.obs[var_name] = new.obs[var_name].astype("bool")
# get all non bool object columns and cast the to category dtype
object_columns = list(new.obs.select_dtypes(include="object").columns)
new.obs[object_columns] = new.obs[object_columns].astype("category")
logg.info(f"The original categorical values `{categorical_names}` were added to obs.")
def _add_categoricals_to_uns(original: AnnData, new: AnnData, categorical_names: list[str]) -> None:
"""Add the original categorical values to uns.
Args:
original: The original AnnData object
new: The new AnnData object
categorical_names: Name of each categorical column
"""
is_initial = "original_values_categoricals" in original.uns.keys()
new["original_values_categoricals"] = {} if not is_initial else original.uns["original_values_categoricals"].copy()
for idx, var_name in enumerate(original.var_names):
if is_initial and var_name in new["original_values_categoricals"]:
continue
elif var_name in categorical_names:
# keep numerical dtype when writing original values to uns
if var_name in original.uns["numerical_columns"]:
new["original_values_categoricals"][var_name] = original.X[::, idx : idx + 1].astype("float")
else:
new["original_values_categoricals"][var_name] = original.X[::, idx : idx + 1].astype("str")
logg.info(f"The original categorical values `{categorical_names}` were added to uns.")
def _check_anndata_input_type(
autodetect: bool | dict, encodings: dict[str, dict[str, list[str]]] | dict[str, list[str]] | str | None
) -> None:
"""
Check type of passed parameters, whether they match the requirements to encode an AnnData object or not.
Args:
autodetect: Whether columns, that should be encoded, should be autodetected or not
Returns:
Whether they match type requirements or not
"""
if not isinstance(autodetect, bool):
raise ValueError(
f"Attempted to encode an AnnData object, but passed parameter for autodetect {autodetect} is not a boolean."
f"Please provide a boolean value for autodetect when encoding a single AnnData object!"
)
elif isinstance(encodings, str) and not autodetect:
raise ValueError("Passing a string for parameter encodings is only possible when using autodetect=True!")
elif autodetect and not isinstance(encodings, (str, type(None))):
raise ValueError(
"[Setting encode mode when autodetect=True only works by passing a string (encode mode name) or None not {type(encodings)}!"
)
class AlreadyEncodedWarning(UserWarning):
pass
class EncodingInputValueError(ValueError):
pass
class AnnDataCreationError(ValueError):
pass
class DuplicateColumnEncodingError(ValueError):
pass
class HashEncodingError(Exception):
pass