test_norm.py

import numpy as np
# import scanpy.api as sc
import pandas as pd

def norm(x, reverse=False):
    if reverse:
        y = np.power(10, x) - 1.01
        y = np.around(y).astype(np.int32)
        return y
    else:
        return np.log10(x + 1.01)

def minmax_0_to_1(x, reverse=False, minmax=1):
    if reverse:
        # x -> [0, 1]
        minmax_x = x * minmax
        # minmax_x -> [0, 6]
        return norm(minmax_x, reverse)
    else:
        norm_x = norm(x, reverse)
        minmax_x = norm_x / np.max(norm_x)
        return minmax_x

array = [0, 1, 2, 3, 150, 134]
x = np.asarray(array)
print(minmax_0_to_1(x))
print(minmax_0_to_1(minmax_0_to_1(x), True, np.max(norm(x))))
print("-----------------")


# def normalization(express_data):
#     adata = sc.AnnData(express_data.T.values)
#     sc.pp.normalize_per_cell(adata)
#     sc.pp.log1p(adata)
#     return pd.DataFrame(adata.X.T, columns=express_data.columns.tolist(), index=express_data.index.tolist())

# a = pd.read_csv("./data/true_counts_simulated_dataset1_dropout0.05.csv")

# print(normalization(a))
# print(normalization(a).describe())