mitocheck_data/utils/normalization_utils.py at main · WayScience/mitocheck_data · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
import pathlib
import pandas as pd
import numpy as np
from sklearn.preprocessing import StandardScaler

from load_utils import compile_mitocheck_batch_data, split_data


def get_normalization_scaler(
    norm_pop_path: pathlib.Path, dataset: str = "CP_and_DP"
) -> StandardScaler():
    """
    get normalization scaler from a normalization population

    Parameters
    ----------
    norm_pop_path : pathlib.Path
        path to normalization output in form of mitocheck IDR stream output
    dataset : str, optional
        which dataset columns to load in (in addition to metadata),
        can be "CP" or "DP" or by default "CP_and_DP"

    Returns
    -------
    StandardScaler
        scaler to be used for data normalization
    """
    # get normalization population
    norm_pop_data = compile_mitocheck_batch_data(norm_pop_path, dataset)

    # derive normalization scaler
    _, norm_pop_feature_data = split_data(norm_pop_data, dataset)
    scaler = StandardScaler()
    scaler.fit(norm_pop_feature_data)

    return scaler


def get_normalized_mitocheck_data(
    data: pd.DataFrame, scaler: StandardScaler()
) -> pd.DataFrame:
    """
    get normalized version of mitocheck data

    Parameters
    ----------
    data : pd.DataFrame
        data to be normalized, in form of compiled mitocheck IDR output
    scaler : StandardScaler
        scaler to use for data normalization

    Returns
    -------
    pd.DataFrame
        normalized data
    """

    # normalize features from data
    col_list = data.columns.tolist()
    derived_features = [col_name for col_name in col_list if "P__" in col_name]
    features = data[derived_features].to_numpy()
    features = scaler.transform(features)
    # make features a dataframe so it can be combined with metadata
    features = pd.DataFrame(features, columns=derived_features)

    # replace original features of data with normalized features
    metadata = [col_name for col_name in col_list if "P__" not in col_name]
    metadata = data[metadata]

    return pd.concat([metadata, features], axis=1)